FAQ (v 0.9)

Archive-name: AudioFAQ/pro-audio-faq
Last-modified: 1996/06/25
Version: 0.9

       Frequently Asked Questions (FAQ) file for
			     Version 0.8

	  Edited and compiled by Gabe Wiener (

Many thanks to all who have contributed to the FAQ.  Individual
contributions are credited at the end of their respective sections.
The core FAQ writers are listed below.  The rubric in brackets will be
used to indicate who has written a particular section.

	Scott Dorsey	[Scott]
	Christopher Hicks	[Chris]
	David Josephson	[David]
	Dick Pierce	[Dick]
	Gabe Wiener		[Gabe]

 *  Each author maintains and asserts full legal copyright on his 
 *  contribution to the FAQ.  Compilation copyright (C) 1996 by 
 *  Gabe M. Wiener.  All rights reserved.  Permission is granted for 
 *  non-commercial electronic distribution of this document.  
 *  Distribution in any other form, or distribution as part of any 
 *  commercial product requires permission. Inquire to


Section I - Netiquette

 Q1.1 - What is this newsgroup for?  What topics are appropriate here, and what
        topics are best saved for another newsgroup?
 Q1.2 - Do I have to be a "professional" to post here?
 Q1.3 - I need to ask the group for help with selecting a piece of equipment.
        What information should I provide in my message?

Section II - The business of audio

 Q2.1 - How does one get started as a professional audio engineer?
 Q2.2 - Are audio schools worth the money?  Which schools are best?
 Q2.3 - What are typical rates for various professional audio services?

Section III - Audio Interconnections

 Q3.1 - How are professional transmission lines and levels different from 
        consumer lines and levels?  What is -10 and +4?  What's a balanced
	or differential line?
 Q3.2 - What is meant by "impedance matching"?  How is it done?  Why is it
 Q3.3 - What is the difference between dBv, dBu, dBV, dBm, dB SPL, and plain
        old dB?  Why not just use regular voltage and power measurements?
 Q3.4 - Which is it for XLRs?  Pin 2 hot?  Or pin 3 hot?
 Q3.5 - What is phantom power?  What is T-power?
 Q3.6 - How do I interconnect balanced and unbalanced components?
 Q3.7 - What are ground loops and how do I avoid them?
 Q3.8 - What is the "Pin 1 problem" and how do I avoid it?
Section IV - Analog tape recording

 Q4.1 - What does it mean to "align" a tape machine? 
 Q4.2 - What is bias?  What is overbias?
 Q4.3 - What is the difference between Dolby A, B, C, S, and SR?  How do each
        of these systems work?
 Q4.4 - What is Dolby HX-Pro?
 Q4.5 - How does DBX compare to Dolby?  
 Q4.6 - How much better are external microphone preamplifiers than those 
        found in my portable recorder?
 Q4.7 - What is an MRL?  Where do I get one?

Section V - Digital recording and interconnection

 Q5.1 - What is sampling?  What is a sampling rate?
 Q5.2 - What is oversampling?
 Q5.3 - What is the difference between a "1 bit" and a "multibit" converter?
	What is MASH?  What is Delta/Sigma?  Should I really care?
 Q5.4 - On an analog recorder, I was always taught to make sure the signal
        averages around 0 VU.  But on my new DAT machine, 0 is all the way at
        the top of the scale.  What's going on here?
 Q5.5 - Why doesn't MiniDisc or Digital Compact Cassette sound as good as DAT
         or CD?  After all, they're both digital.
 Q5.6 - What is S/P-DIF?  What is AES/EBU?
 Q5.7 - What is clock jitter?
 Q5.8 - How long can I run AES/EBU or S/P-DIF cables?  What kind of cable
	should I use?
 Q5.9 - What is SCMS?  How do I defeat it?
 Q5.10 - What is PCM-F1 format?
 Q5.11 - How do digital recorders handle selective synchronization?
 Q5.12 - How can a 44.1 kHz sampling rate be enough?
 Q5.13 - Doesn't the 44.1 kHz sampling rate make it impossible to 
         reproduce square waves? 
 Q5.14 - How can a 16-bit word-length be enough?
 Q5.15 - What's all this about 20- and 24-bit digital audio?  Aren't
         CDs limited to 16 bits?

Section VI - Digital editing and mastering

 Q6.1 - What is a digital audio workstation?
 Q6.2 - How is digital editing different from analog editing?
 Q6.3 - What is mastering? 
 Q6.4 - What is normalizing?
 Q4.5 - I have a fully edited DAT that sounds just like I want it to sound on
        the CD.  Is it okay to send it to the factory?
 Q6.6 - What is PCM-1630?  What is PMCD?
 Q6.7 - When preparing a tape for CD, how hot should the levels be?
 Q6.8 - Where can I get CDs manufactured?
 Q6.9 - How are CD error rates measured, and what do they mean?

Section VII - Market survey.  What are my options if I want --

 Q7.1 - A portable DAT machine
 Q7.2 - A rack size DAT machine
 Q7.3 - An inexpensive stereo microphone
 Q7.4 - An inexpensive pair of microphones for stereo
 Q7.5 - A good microphone for recording vocals
 Q7.6 - A good microphone for recording [insert instrument here]
 Q7.7 - A a small mixer
 Q7.8 - A portable cassette machine
 Q7.9 - A computer sound card for my IBM PC or Mac
 Q7.10 - An eight-track digital recorder?

Section VIII - Sound reinforcement

 Q8.1 - We have a fine church choir, but the congregation can't hear them.
        How do we mic the choir?
 Q8.2 - How do I 'ring out' a system?
 Q8.3 - How much power to I need for [insert venue here]?
 Q8.4 - How good is the Sabine feedback eliminator?

Section IX - Sound restoration

 Q9.1 - How can I play old 78s?  
 Q9.2 - How can I play Edison cylinders?
 Q9.3 - What are "Hill and Dale" recordings, and how do I play them back?
 Q9.4 - What exactly are NoNOISE and CEDAR?  How are they used?
 Q9.5 - How do noise suppression systems like NoNOISE and CEDAR work?
 Q9.6 - What is forensic audio?  

Section X - Recording technique, Speakers, Acoustics, Sound

 Q10.1 - What are the various stereo microphone techniques?
 Q10.2 - How do I know which technique to use in a given circumstance?
 Q10.3 - How do I soundproof a room?
 Q10.4 - What is a near-field monitor?
 Q10.5 - What are the differences between "studio monitors" and home

Section XI - Industry information

Q11.1 - Is there a directory of industry resources?
Q11.2 - What are the industry periodicals?
Q11.3 - What are the industry trade organizations?
Q11.4 - Are there any conventions or trade shows that deal specifically 
        with professional audio?

Section XII - Miscellaneous

 Q12.1 - How do I modify Radio Shack PZMs?
 Q12.2 - Can I produce good demos at home?
 Q12.3 - How do I remove vocals from a song?

Section XIII - Bibliography

 Q13.1 - Fundamentals of Audio Technology
 Q13.2 - Studio recording techniques
 Q13.3 - Live recording techniques
 Q13.4 - Digital audio theory and practice
 Q13.5 - Acoustics 
 Q13.6 - Practical recording guides

Section XIV

 Q14.1 - Who wrote the FAQ
 Q14.2 - How do you spell and pronounce the FAQ maintainer's surname?


Section I - Netiquette

Q1.1 - What is this newsgroup for?  What topics are appropriate here, and what
       topics are best saved for another newsgroup?

  This newsgroup exists for the discussion of issues and topics related
  to professional audio engineering.  We generally do not discuss issues
  relating to home audio reproduction, though they do occasionally come
  up.  The* hierarchy of newsgroups is as follows:		Issues pertaining to professional audio	Buying and trading of consumer equipment 		Technical discussions about consumer audio	Everyone's $0.02 on consumer audio	High-end consumer audio discussions		Everything else

  Please be sure to select the right newsgroup before posting.

Q1.2 - Do I have to be a "professional" to post here?

  No.  Anyone is welcome to post on so long as the messages
  you post are endemic to the group in some way.  If you are not an audio
  professional, we would ask that you read this FAQ in full before posting.
  You may find that some of your essential questions about our field are
  answered right here.  But if not, feel free to ask us.

Q1.3 - I need to ask the group for help with selecting a piece of equipment.
       What information should I provide in my message?

  If you are going to post a request for advice on buying equipment,
  please provide the following information.

	Your application for the equipment
	What other equipment you will be using it with
	Your budget for the equipment
	Any specific requirements the equipment should have

  There is nothing worse than messages like "Can anyone recommend a DAT
  machine for me to buy???"  Sure we can.  But what do you want to _do_ 
  with it?  We can recommend DAT machines for $400 or for $14,000.

Section II - The business of audio

Q2.1 - How does one get started as a professional audio engineer?

  There are as many getting-started stories as there are audio
  engineers.  The routes into the industry are highly dependent on what
  aspect of the industry one wishes to enter.  For instance, many
  engineers who work in the classical-music field have at one time or
  another been classical performers.  Others enter through their work in
  other musical genres, or through engineering programs at universities
  or technical schools.  Without exception, everyone in the industry has
  learned at least a portion of their craft from watching those with
  more hands-on experience.  Whether this comes from a formal internship
  or just from sustained observation and long-term question-asking, it
  is almost always universally true. [Gabe]

Q2.2 - Are audio schools worth the money?  Which schools are best?

  An audio school will teach you the basics of the audio business, 
  but just like any technical school, what they teach you may not be 
  worth what you pay.

  There are several schools of thought:

  1. Audio schools are great, you get trained on the gear that is used
     by top studios and costs millions of dollars, you get taught by pros
     in the field and you have job placement assistance after you graduate.

  2. Going to an audio school is like wanting to learn aviation,
     and when you start flight school they teach you a 747.  In the
     real world, you are probably not going to have 96 channel automated
     consoles on your first job.  You are not going to mix your first 
     live gig on a 48-channel 100,000 watt stadium PA rig.  Better to 
     start off on real world equipment and work your way up to the 
     top-of-the-line stuff.  Most recording studios are 24-track analog 
     or less and most PA systems are 16 channel, 3,000 watts or less.  
     Don't buy education for something you will never get to use after 
     you leave the school.

  3. Audio Schools are a waste of money.  Instead of spending $18,000 
     for a course and having nothing to show for it but a technical 
     certificate (which everyone knows is no help at all getting a job), 
     you would be better off spending the 18 grand on books and gear and 
     learning by trial and error, or saving the 18 grand altogether and 
     learning first from reading, and later from apprenticing.
	[ (Jim Saurman)]

  Jim summarizes the opinions pretty well.  Recognize that an altogether
  different option is to attend a full four-year college program.  Many
  colleges and universities offer such programs.  Examples include 
  Peabody Conservatory, Cleveland Institute of Music, McGill University,
  New York University, University of Miami at Coral Gables, and the
  University of Massachusetts at Lowell.  Without fail, graduates from
  these sorts of programs earn far more respect than graduates of any
  technical school.  [Gabe]

Q2.3 - What are typical rates for various professional audio services?

  Depends on what you want to have done, and where.

  One can pay upwards of $300/hr for prime studio rental time in New York.
  In a small community however, one might find a project studio for $25/hr.
  Generally speaking, the rule is: the rarer the service, the more it will
  cost.  In a community with dozens of small 8-track studios, you won't
  have to pay much.  If you need emergency audio restoration, or mastering
  by a top-flight pop-music engineer, you can expect to drop many hundreds
  of dollars an hour.  Like so many other things in this industry, there
  are no rules, and Smith's invisible hand guides the market. [Gabe]

Section III - Audio Interconnections

Q3.1 - How are professional transmission lines and levels different from 
       consumer lines and levels?  What is -10 and +4?  What's a balanced
       or differential line?

  Professional transmission lines differ from consumer lines in two
  ways.  First, consumer lines tend to run about 14 dB lower in level
  than pro lines.  Second, professional lines run in differential, or
  balanced, configuration.

  In a single-ended line, the signal travels down one conductor and
  returns along a shield.  This is the simplest form of audio
  transmission, since it is essentially the same AC circuit you learned
  about in high-school physics.  The problem here is that any noise
  or interference that creeps into the line will simply get added to
  the signal and you'll be stuck with it.

  In a differential line, there are three conductors.  A shield, a
  normal "hot" lead, and a third lead called the "cold" or "inverting"
  lead, which carries a 180-degree inverted copy of the hot lead.  Any
  interference that creeps into the cable thus affects both the hot and
  cold leads equally.  At the receiving end, the hot and cold leads are
  summed using a differential amplifier, and any interference that has
  entered the circuit (called "common-mode information" since it is
  common to both the hot and cold leads), gets canceled out.
  Differential lines are thus better suited for long runs, or for
  situations where noise or interference may be a factor.  [Gabe]

 Q3.2 - What is meant by "impedance matching"?  How is it done?  Why is it

  We can talk about the characteristic impedance of an input, which is to
  say the ratio of voltage to current that it likes to see, or how much
  it loads down a source.  (You can think of this as being an "AC resistance"
  and you would be mostly right, although it's actually the absolute 
  magnitude of the vector drawn by the resistive and reactive load
  components.  Dealing with line level signals, reactive components
  are going to be negligible, though).

  In general, in this modern world, most equipment has a low impedance
  output, going into relatively high impedance input.  This wastes some
  amount of power, but because electricity is cheap and it's possible to
  build low-Z outputs easily today, this is not a big deal.

  With microphones, it _is_ a big deal, because the signal levels are
  very low, and the drive ability poor.  As a result, we try and get the
  best efficiency possible from microphones to get the lowest noise
  floor.  This is often done by using transformers to step up the voltage
  or step it down, to go into a higher or lower Z load.  Transformers
  have some major disadvantages in that they can be significant sources
  of nonlinearity, but back in the days of tubes they were the only
  solution.  Tubes have a very high-Z input, and building balanced inputs
  with tubes requires three devices instead of one.  As a result, all
  mike preamps would have a 600 ohm balanced input, with a transformer,
  driving a preamp tube.  Today, transistor circuits can be used for 
  impedance matching, although they are often more costly and can be noisier
  in cases.

  As a result of the expense, consumer equipment was built with high-Z 
  microphone inputs, and high-Z microphones.  This resulted in more noise
  pickup problems, but was cheaper to make.  Unfortunately this still
  held on into the modern day of the transistor, and a lot of high-Z
  consumer gear exists.  Guitar pickups are generally high-Z devices,
  and require a direct box to reduce the impedance so that they can go into
  a standard 600 ohm mike preamp directly.

  Many years ago, the  techniques that were used in audio came originally
  from telephone company practice.  Phone systems operate with 150 or 600
  ohm balanced lines, and adoption of this practice into the audio industry
  caused those standards to be used.  In the modern age where lines are
  relatively short and transformers considered problematic, the tendency
  has been to have low-Z outputs for all line level devices,  driving
  high-Z inputs.  While this is not the most efficient system, it is relatively
  foolproof, and appears on most consumer equipment.  A substantial amount of
  professional gear, however, still uses internal balancing transformers or
  resistor networks to match to a perfect 600 ohm impedance.   [Scott]

  [Ed. note: Modern equipment works on principles of voltage transfer
  rather than power transfer.  Thus a standard audio circuit today is
  essentially a glorified voltage divider.  You have a very low output
  impedance and a very high input impedance such that the most voltage
  is dropped across the load.  This is not an impedance-matched circuit
  in the classic sense of the word.  Rather, it is a "bridged" or
  "constant voltage" impedance match, and is the paradigm on which
  nearly all audio circuits operate nowadays. -Gabe]

Q3.3 - What is the difference between dBv, dBu, dBV, dBm, dB SPL, and plain
       old dB?  Why not just use regular voltage and power measurements?

  Our ears respond logarithmically to increases in sound pressure level.
  In order to simplify the calculations of these levels, as well as the
  electrical equivalents of them in audio systems, the industry uses a
  logarithmic system to denote the values.  Specifically, the decibel is
  used to denote logarithmic level above a given reference.  For
  instance, when measuring sound pressure level, the basic reference
  against which we take measurements is the threshold of hearing for
  the average individual, 10^-12 W/m^2.  The formula for dB SPL then

	10 Log X / 10^-12  where X is the intensity in W/m^2

  The first people who were concerned about transmitting audio over
  wires were, of course, the telephone company.  Thanks to Ma Bell we
  have a bunch of other decibel measurements.  We can use the decibel to
  measure electrical power as well.  In this case, the formula is
  referenced to 1 milliwatt in the denominator, and the unit is dBm.  1
  milliwatt was chosen as the canonical reference by Ma Bell.  Since
  P=V^2 / R, we can also express not only power gain in dB but also
  voltage gain.  In this case the equation changes a bit, since we have
  the ^2 exponent.  When we take the logarithm, the exponent comes
  around into the coefficient, making our voltage formula 20 log.  
  In the voltage scenario, the reference value becomes 0.775 V (the
  voltage drop across 600 ohms that results in 1 mW of power).  The
  voltage measurement unit is dBv.

  The Europeans, not having any need to abide by Ma Bell's choice for a
  canonical value, chose 1V as their reference, and this is reflected
  as dBV instead of dBv.  To avoid confusion, the Europeans write the
  American dBv as dBu.  Confused yet?  [Gabe]

Q3.4 - Which is it for XLRs?  Pin 2 hot?  Or pin 3 hot?

  Depends on whom you ask!  Over the years, different manufacturers have
  adopted varying standards of pin 2 hot and pin 3 hot (and once in a
  while, pin *1* hot!).  But nowadays most manufacturers have adopted  
  pin 2 hot.  Still, it is worth taking the extra minute or two to check  
  the manual.  The current AES standard is pin 2 hot.  [Gabe]

Q3.5 - What is phantom power?  What is T-power?

  Condenser microphones have internal electronics that need power
  to operate. Early condenser microphones were powered by
  batteries, or separate power supplies using multi-conductor
  cables. In the late 1960's, German microphone manufacturers
  developed 2 methods of sending power on the same wires that carry
  the signal from the microphone.

  The more common of these methods is called "phantom power" and is
  covered by DIN spec 45596. The positive terminal of a power
  supply is connected through resistors to both signal leads of a
  balanced microphone, and the negative terminal is connected to
  ground. 48 volts is the preferred value, with 6800 ohm resistors
  in each leg of the circuit, but lower voltages and lower resistor
  values are also used. The precise value of the resistors is not
  too critical, but the two resistors must be matched within 0.4%.

  Phantom power has the advantage that a dynamic or ribbon mic may
  be plugged in to a phantom powered microphone input and operate
  without damage, and a phantom powered mic can be plugged in to
  the same input and receive power. The only hazard is that in case
  of a shorted microphone cable, or certain old microphones having
  a grounded center tap output, current can flow through the
  microphone, damaging it. It's a good idea anyway to check cables
  regularly to see that there are no shorts between any of the
  pins, and the few ribbon or dynamic microphones with any circuit
  connection to ground can be identified and not used with phantom

  T-power (short for Tonaderspeisung, also called AB or parallel
  power, and covered by DIN spec 45595) was developed for portable
  applications, and is still common in film sound equipment.
  T-power is usually 12 volts, and the power is connected across
  the balanced pair through 180 ohm resistors. Only T-power mics
  may be connected to T-power inputs; dynamic or ribbon mics may be
  damaged and phantom powered mics will not operate properly. [David]

Q3.6 - How do I interconnect balanced and unbalanced components?

  First, let's define what the terms mean. The simplest audio
  circuit uses a single wire to carry the signal; the return path,
  which is needed for current to flow in the wire, is provided
  through a ground connection, usually through a shield around the
  wire.  This system, called unbalanced transmission, is very
  susceptible to hum pickup and cannot be used for low level
  signals, like audio, for more than a few feet. Balanced
  transmission occurs when two separate and symmetrical wires are
  used to carry the signal. A balanced input is sensitive only to
  voltage that appears between the two input terminals; signals
  from one terminal to ground are canceled by the circuit.

  The simplest way to connect between balanced and unbalanced
  equipment is to use a transformer. The signals are magnetically
  coupled through the core of the transformer and either side may
  be balanced or unbalanced. Good transformers are expensive,
  however, and there are cheaper methods that can be used in some

  An unbalanced output can be connected to a balanced input. For
  instance, from the unbalanced output of a CD player, connect the
  center pin to pin 2 of the balanced XLR input connector, and the
  ground to pins 1 and 3.  To connect the balanced output of something
  to an unbalanced input requires different techniques depending on
  whether the output is active balanced (each side has a signal with
  respect to ground) or floating balanced (for instance, the secondary
  of a transformer with no center-tap connection). If it's an active
  balanced output, you can simply use half of it; connect pin 2 to the
  unbalanced input, and pin 1 to ground, leaving pin 3 floating. If this
  doesn't work (no or very weak signal) connect pin 3 of the output to 
  pin 1 and ground and leave pin 2 connected to the unbalanced input 
  center pin. Some active balanced outputs, particularly microphones, 
  use the balanced circuit to cancel distortion, so this hookup may
  result in higher distortion than if a proper balanced-to-unbalanced
  converter such as a differential stage or a transformer were used.

Q3.7 - What are ground loops and how do I avoid them?

  One of the most difficult troubleshooting tasks for the audio
  practitioner is finding the source of hum, buzz and other
  interfering signals in the audio signal. Often these are caused
  by "ground loops." This unfortunate and inaccurate term (it need
  not be in the "ground" path, and the "loop" is not what causes
  the problem) is poorly understood by most users of audio
  equipment. A better name for this phenomenon is "shared path
  coupling" because it happens when two signals share the same
  conductor path and couple to each other as a result.

  Another semantic problem that should be addressed early on is the
  idea that "ground" is one place where all currents go. It's not,
  there's nothing special about calling a signal "ground," current
  still flows through any path that's available to it.

  Referring to the discussion above regarding unbalanced signal
  paths, recall that there must be a complete circuit from the
  output of some device, through the input of another device and
  back to the "return" side of the output if any current is to
  flow. Current doesn't flow by itself, it must have a complete
  path. If there are multiple paths over which the current might
  flow, the current will be divided among them with most of the
  current flowing through the path having the least resistance. Any
  available path, regardless of the resistance in it, will carry
  some of the current, it's not a case of all the current following
  the path that has least resistance.

  For example, suppose we have two units connected together through
  a small piece of coaxial cable, and the units are also connected
  together at the wall outlet through their grounded power cords --
  the ground pins are connected to the chassis at each end. The
  audio signal goes along the center of the coaxial cable, and part
  of it might come back along the shield of the coax, but part will
  also go through the ground wire of one unit and back through the
  ground wire of the other unit. A problem arises when some other
  signal is also flowing through this same return path. The other
  signal might be another audio signal, video, data, or power. All
  of the currents in a wire add together, and the resistance of the
  wire causes a voltage to appear in proportion to the current
  flowing. All of these voltages add together, so there is a little
  bit of the video signal added to the audio, some of the power
  signal added to the video, some of the power signal added to the
  audio, etc. In rare instances, the "loop" of wire formed by the
  intended ground return path and the happenstance lower resistance
  return path formed by mounting hardware, power cords, etc. can
  form a magnetic pickup as well, so that magnetic fields radiated
  by transformers, CRT's, etc. can also induce a current in the
  "loop," which makes yet another source of noise voltage.

  This shared path coupling is a constant problem with unbalanced
  audio systems. Lots of different methods have been tried to get
  around the problem, many of them dangerous. Clipping off the
  ground leads of equipment so there is no common power line path
  between them simply makes any fault or leakage current follow
  some other path, back through the signal cable to some equipment
  that has a ground -- perhaps through the user's body, if all the
  ground pins have been removed. The only general solution to
  "ground loop" coupling with unbalanced equipment is to connect
  all the chassis together with a very low resistance path (copper
  strap or braid, for example), on the principle that since the
  resistance is so low, any leakage current will produce a
  correspondingly low signal voltage. It may also be effective to
  interrupt the ground path of shield conductors over signal wires;
  force the return path to go through the designated common strap
  while leaving the shield in place only for electrostatic

  With balanced equipment, no current should be flowing in the
  shield conductors, and in fact performance should be identical
  with the shield left disconnected at one end (preferably the
  receiver end). Therefore balanced systems should be impervious to
  shared path coupling or "ground loop" problems but in fact they
  aren't, because most signals inside a given piece of equipment
  are unbalanced, and there are often return paths internal to the
  equipment that can be shared with return paths between other
  units of equipment connected to it. Especially with mixed
  digital, video and audio signals and high gain, high negative
  feedback amplifier circuitry, this can be a big problem -- small
  currents can create big effects -- and this brings us to the next
  question.  [David]

Q3.8 - What is the "Pin 1 problem" and how do I avoid it?

  This is a special case of "ground loop" or shared path coupling.
  Recently this has been discussed in great detail and clarity by a
  group led by the consultant Neil Muncy of Toronto. Suppose you
  have a mixer, whose balanced output is connected to an
  amplifier's balanced input through a correctly wired cable. Both
  units are powered from the AC mains and one or both have some
  small amount of AC leakage current that travels to ground through
  all available ground paths -- including the shield of the cable
  that connects the two units. So far so good, no harm done because
  the circuit is balanced and any common mode voltage from current
  flowing through the shield will be canceled by the amplifier
  input. However... a small part of this leakage current also
  travels through the shield of the wire going from the back panel
  XLR connector to the PC board, through some "ground" traces on
  the PC board, and back out through the power line ground cable.
  No problem so far, except that some gain stage on that same PC
  board also uses that piece of ground trace in its negative
  feedback loop, and some part of that leakage signal will be added
  to the signal in that gain stage; it might be video, or data, or
  another audio signal, or (most commonly) power.

  The solution to this variant of shared path coupling is the same
  sort of approach that applies to other unbalanced signals: give
  the leakage current a very low resistance path to follow, and
  remove as many of the shared paths as possible. Within a unit of
  equipment, all the XLR connectors' pin 1 terminals should be
  connected to ground with very low resistance (big) wire or
  traces, and preferably all of the ground connections should be
  made at one point, the so-called "star ground" system. A brute
  force approach is to assume that the back panel is the star
  ground, and wire every connector's pin 1 solidly to the panel as
  directly as possible, and lift all the ground wires but one that
  go from the connectors to the circuitry. In this way, all the
  external leakage currents (the "fox" to use Neil Muncy's term)
  will be conducted through the back panel and out of the way,
  rather than running them through the ground traces on the PC
  board where they will mix with internal low level signals in high
  gain stages (the "hen house"). Individual wires can be run from
  points on the circuit board that need to be at "ground" potential
  to a common point on the back panel, which is designated a "zero
  signal reference point" (ZSRP). Equipment that has a reputation
  for being "quiet" and easy to use in many different applications
  is often found to be wired this way, while equipment that is
  "temperamental" if often found to be wired in such a way that
  leakage currents are easily coupled to internal signal lines.

  There's a simple test that can be done to check equipment
  susceptibility to this problem. Connect the output, preferably
  balanced and floating, of an ordinary audio oscillator to the pin
  1 of any two XLR connectors on the equipment. Now operate the
  equipment through its various modes, gain settings, etc. You may
  be surprised to find the audio oscillator's signal appearing in
  many different places in the equipment. [David]

Section IV - Analog tape recording

Q4.1 - What does it mean to "align" a tape machine? 

  There are a number of standard adjustments on any analogue tape
  machine, which can roughly be broken up into mechanical and electronic
  adjustments.  The mechanical adjustments include the head position
  (height, skew, and azimuth), and sometimes tape speed.  Incorrect head
  height will result in poor S/N and leakage between channels, because
  the tracks on the head do not match up exactly with those on the tape.
  Incorrect tape skew will result in level differences between channels
  and uneven head wear, because there is more pressure on the top of the
  head than the bottom (or vice versa).  Incorrect azimuth will result
  in loss of high frequency response and strange skewing of the stereo
  image.  Tape speed error will result in tonal shifts, although on many
  machines with capstan speed controlled by crystal or line frequency,
  it is not adjustable.

  Electronic adjustments include level and bias adjustments for each
  channel.  Some machines may have bias frequency adjustments, equalization
  adjustments for playback and record emphasis, pre-distortion adjustments,
  and a varied bevy of adjustments for noise reduction systems.

  Alignment is relatively simple, and the same general method applies
  from the smallest cassette deck to the largest multitrack machine.
  First, put a test tape on the machine.  Use a real reference tape,
  from the manufacturer, from MRL, or a similarly legitimate lab.  DO
  NOT EVER use a homebrew test tape that was recorded on a "known good"
  machine.  You will regret it someday.  Spend the money and get a real
  test tape (and not one of the flaky ones from RCA).

  1. Speed adjustment (if necessary).  Play back a 1 KHz reference tone
     and, using a frequency counter, adjust the tape speed for proper
     frequency output.  There are strobe tapes available for this as well,
     but with cheap frequency counters available, this method is much easier.

  2. Head height and skew adjustments.  Better see your machine's manual
     on this one, because I have seen a variety of ways of doing this.

  3. Azimuth adjustment.  I find the easiest way to do this is to take
     the left and right outputs and connect them to the X and Y inputs
     of an oscilloscope, and play back a 1 KHz reference tone, while
     adjusting the azimuth until a perfectly-diagonal line appears.
     You can do this by ear if you are desperate, but I strongly recommend
     the lissajous method, which is faster and more accurate.  On multitrack
     decks, use the two tracks as close as possible to the edge of the tape.
     Now you have the playback head azimuth set... put a 1 KHz source into
     the record input, with a blank tape on the machine, and adjust the
     azimuth of the record head for the proper diagonal line.

  4. Playback eq adjustment (if necessary).  This is a case of playing
     back various test tones at different frequencies, and adjusting the
     response curve of the deck to produce a flat output.  You can also
     do this by playing back white noise and using a third-octave spectrum
     analyzer of great accuracy to adjust for flat response.  Again, this
     is one to check your deck's manual for, because the actual 
     adjustments vary from one machine to another, and you will want to 
     use the test tape once again.

  5. Record eq adjustment (if necessary).  How this is done (and whether you
     want to do it after biasing the tape) depends a lot on your deck.

  6. Bias adjustment.  There are a lot of ways to do this.  My favorite method
     is to use a white noise source, and adjust the bias until the source and
     tape output sound identical.  Some people prefer to use a signal generator
     and set so that the levels of recorded tones at 1 KHz and 20 KHz are
     identical.  I find I can get within .5 dB by ear, though your mileage
     may differ.  [Ed. note:  Many tapes have recommended overbias settings,
     and many decks will also provide a chart that correlates the amount of
     overbias against available tape formulations.  -Gabe]

  7. Record level adjustment.  I use a distortion analyzer, and set the level
     so that at +3 dB, I get 3% distortion on the output.  Some folks who are
     using very hot tape set the machines so that a certain magnetic flux is
     produced at the heads given a certain input, but I find setting for
     a given distortion point does well for me.  If you don't have a distortion
     analyzer, use a 1 KHz tone source and set so that you have the onset of
     audible distortion at +3 dB, and you will be extremely close.

     [Ed. note:  The traditional way to do this is to align the repro side
     of the machine using a calibration tape, and then to put the machine
     into record.  Monitoring off the repro head, the operator then aligns
     the record electronics until the output is flat. -Gabe]

  At this point, you will be pretty much set.  Whether you want to do this
  all on a regular basis is a good question.  You should definitely go
  through the complete procedure if you ever change brands of tape.  Checking
  the mechanical parameters on a regular basis is a good idea with some
  decks (like the Ampex 350), which tend to drift.  Clean your heads
  every time you put a new reel on, and demagnetize regularly.  [Scott]

Q4.2 - What is bias?  What is overbias?

  With just the audio signal applied to a tape, the frequency response
  is very poor.  High frequency response is much better than low
  frequency, and the low frequency distortion is very high.  In 1906,
  the Poulson Telegraphone managed to record an intelligible voice on a
  magnetic medium, but it was not until the 1930s when this problem was
  solved by German engineers.  

  To compensate for the tape characteristic, a very high frequency
  signal is applied to the tape in addition to the audio.  This is
  typically in the 100 KHz range, far above the audio range.  With the
  bias adjusted properly, the frequency response should be flat across
  the audible range.  With too low bias, bass distortion will be the
  first audible sign, but with too much bias, the high frequency
  response will drop off.

  Incidentally, digital recording equipment takes advantage of the very
  nonlinearity that is a problem with analogue methods.  It records a
  square wave on the tape, driving the tape into saturation at all
  times, and extracts the signal from the waveform edges.  As a result,
  no bias is required.  (For a good example of the various digital
  recording methods, check out NASA SP 5038, _Magnetic Tape Recording_.)

  [Ed. note: For those looking for an understanding of why we need
  bias in the first place, here is one way to think about it.  Tape
  consists of lots of small magnetic particles called domains.  These
  domains are exposed to a magnetic field from the record head and 
  oscillate in polarity as the AC signal voltage changes.  Domains,
  being physical objects, have inertia.  Every time the analog signal
  crosses from positive to negative and back again, the voltage passes
  the zero point for an instant.  At this moment, the domain is at rest,
  and like any other physical object, there is a short period of inertia
  before it gets moving again.  The result is the bizarre high-frequency
  performance characteristic that Scott described.  The high frequency of
  a bias signal simply ensures that the domains are always kept in motion,
  negating the effect of inertia at audio frequencies.  -Gabe]

Q4.3 - What is the difference between Dolby A, B, C, S, and SR?  How do each
        of these systems work?

  The Dolby A, B, C, SR, and S noise reduction (NR) systems are non-linear
  level-dependent companders (compressors/expanders). They offer various
  amounts of noise reduction, as shown in the table below.

    Dolby   HF NR   LF NR  Number Of Active              Target
    System  Effect  Effect Frequency Bands               Market     Year
    ------  ------  ------ ----------------------------  ---------  ----
      A     10 dB   10 dB  4 fixed                       Pro audio  1967
      B     10 dB   --     1 sliding (HF)                Domestic   1970
      C     20 dB   --     1 sliding (HF)                Domestic   1981
      SR    24 dB   10 dB  1 sliding (HF), 1 fixed (LF)  Pro audio  1986
      S     24 dB   10 dB  1 sliding (HF), 1 fixed (LF)  Domestic   1990
    ------  ------  -----  ----------------------------  ---------  ----

  The band-splitting system used with Dolby A NR is a relatively costly
  technique, although it can deal with noise at all frequencies. The
  single sliding band techniques used in Dolby B and C systems are less
  costly, making them more suitable for consumer tape recording
  applications where the dominant noise contribution occurs at high

  The typical on-record frequency response curves for the Dolby B NR
  system look something like those depicted below. The curves for Dolby C,
  SR, and S are similar, but the actual response levels and behaviour at
  high frequencies are modified to extract better performance form these
  more advanced systems.

    0dB -|----------------------------------------------------
  -10dB -|
         |                    /-------------------------------
         |                   /
  -20dB -|------------------/
  -30dB -|                      /-----------------------------
         |                    /
         |                  /
  -40dB -|----------------/
         |                   |                                 |
        20Hz                1kHz                             20kHz

  The above picture attempts to show that the encoding process provides
  selective boost to high frequency signals (decoding is the exact
  reciprocal), and the curves correspond to the results achieved when no
  musical signal is applied. The amount of boost during the compansion
  depends on the signal level and its spectral content. For a tone at
  -40dB at 3 kHz, the boost applied to signals with frequencies above this
  would probably be the full 10dB allowed by the system. If the same tone
  were at a level of -20dB, then the boost would be less, maybe about 5dB.
  If the tone was at 0dB, then no boost would be supplied, as tape
  saturation would be increased (beyond it's normal amount).

  The single band of compansion utilized with Dolby B NR reaches
  sufficiently low in frequency to provide useful noise reduction when no
  signal is present. Its width changes dynamically in response to the
  spectral content of music signals. As an example, when used with a solo
  drum note the companding system will slide up in frequency so that the
  low frequency content of the drum will be passed through at its full
  level. On replay, the playback of the bass drum is allowed to pass
  through without modification to its level, while the expander lowers the
  volume at high frequencies above those of the bass drum, thus providing
  a reduction in tape hiss where there is no musical signal. If a guitar
  is now added to the music signal, the companding band slides further up
  in frequency allowing the bass drum and guitar signals through without
  any compansion, while still producing a worthwhile noise reduction
  effect at frequencies above those of the guitar.

  The Dolby B NR system is designed to start taking effect from 300Hz, and
  its action increases until it reaches a maximum of 10dB upwards of 4kHz.
  Dolby C improves on this by taking effect from 100Hz and providing about
  15dB of NR at 400Hz, increasing to a maximum of 20dB in the critical
  hiss region from 2kHz to 10kHz. Dolby C also includes spectral skewing
  networks which introduce a rolloff above 10kHz prior to the compander
  when in encoding mode. This helps to reduce compander errors caused by
  unpredictable cassette response above 10kHz, and an inverse boost is
  added after the expander to compensate. Although this reduces the noise
  reduction effect above 10kHz, the ear's sensitivity to noise in that
  region is diminished, and the improved encode/decode tracking provides
  important improvements in overall system performance. An anti-saturation
  shelving network, beginning at about 2kHz, also acts on the high
  frequencies but it only affects the high-level signals that would cause
  tape saturation. A complementary network is provided in the decode chain
  to provide overall flat response.

  When the tape is played back, the inverse of the above process takes
  place. For an accurate decoding to occur, it is necessary that playback
  takes place with no offsets in levels between record and replay. i.e. If
  a 400 Hz tone is recorded at 0dB (or -20dB), then it must play back at
  0dB (or -20dB). This will help ensure correct Dolby "tracking".

  Just think about it: if a -40dB tone at 8kHz was recorded with
  Dolby B on, then it would actually have a level of -30dB on tape.
  The same tone, if it were at a -20dB level, would have a level of
  about -15dB on tape. If the sensitivity of the tape was such
  that anything recorded at 0dB actually went on tape as -10dB,
  then you can see that the Dolby encoded tones would actually be
  at a lower level, and the system would have no way of determining
  this. It assumes 0dB in = 0dB out. Hence the signal would be
  decoded with the incorrect amount of de-boost.

  The Dolby SR and S NR systems provide slightly more NR than Dolby C at
  high frequencies, 24dB vs 20dB, but they also achieve a 10dB NR effect
  at low frequencies below 200Hz as well. This is obtained using a
  two-band approach, the low-frequencies being handled by a fixed-band
  processor, while the high frequencies are tackled by a sliding band
  processor. This reduces the potential for problems such as "noise
  pumping", caused by high-level low frequency transient signals (bass
  notes from drums, double basses, organs), raising the sound level in a
  cyclic fashion. Dolby SR and S also contain the spectral skewing and
  anti-saturation circuits for high-level high-frequency signals that are
  implemented with Dolby C. The performance of the sliding band is
  improved over that obtained with Dolby B and C NR systems by reducing
  the degree of sliding that occurs in the presence of high-frequency
  signals. This increases the noise reduction effect available at
  frequencies below those occurring in the music signal.

  An additional benefit of the Dolby S NR system for consumers is that the
  manufacturers of cassette decks who are licensed to use the system must
  adhere to a range of strict performance standards. These include an
  extended high frequency response, tighter overall response tolerances, a
  new standard ensuring head height accuracy, increased overload margin in
  the electronics, lower wow and flutter, and a head azimuth standard.
  These benefit users by enhancing the performance of cassette recorders
  as well as helping to ensure that tapes recorded on one deck will play
  back accurately on any other. [Witold Waldman -]

Q4.4 - What is Dolby HX-Pro?

  HX-Pro is a scheme to reduce the level of the bias signal when high
  frequency information is present in the recorded signal.  Sufficient
  high frequency information will act to bias the tape itself, and by
  reducing the AC bias signal somewhat, additional signal can be applied
  without saturating the tape.  This is a single-ended system; it
  requires no decoding on playback, because it merely permits more
  signal to be recorded on the tape.  In theory it is an excellent idea,
  and some implementations have lived up to the promise of the method,
  although some other implementations have produced unpleasant
  artifacts. [Scott]

Q4.5 - How does DBX compare to Dolby?  

[Anyone?  Anyone?  -G]

 Q4.6 - How much better are external microphone preamplifiers than those 
        found in my portable recorder?

  Going by the rule that "external is better than internal," the
  external preamps are likely to sound better.  Besides the issue of
  electrical shielding and interaction, it is simply the case that a
  designer who is spending *all* his time on a project designing only a
  preamp is likely to do a better job of it than a tape machine design
  team that has to worry how they're going to fit the preamp into the
  box and still have enough room for the rest of the tape machine. [Gabe]

Q4.7 - What is an MRL?  Where do I get one?

  An MRL is a reference alignment tape from Magnetic Reference Laboratory.
  These tapes, available in every conceivable tape speed, tape width, 
  equalization, and field strength, contain alignment tones useful in
  calibrating the electronics of analog tape machines.

  These tapes can be ordered from many pro audio dealers.  If not, you
  can contact MRL directly at:

	Magnetic Reference Laboratory, Inc.
	229 Polaris Avenue, Suite 4
	Mountain View, CA 94043

	Tel: (415) 965-8187
	Fax: (415) 965-8548

Section V - Digital recording and interconnection

Q5.1 - What is sampling?  What is a sampling rate?

  Sampling can be (roughly) defined as the capture of a continuously
  varying quantity at a precisely defined instant in time. Most usually,
  signals are sampled at a set of sample-points spaced regularly in
  time. Note that sampling in itself implies nothing about the
  representation of sample magnitude by a number. That process is called

  The Nyquist theorem states that in order to faithfully capture all of
  the information in a signal of one-sided bandwidth B, it must be
  sampled at a rate greater than 2B. A direct corollary of this is that
  if we wish to sample at a rate of 2B then we must pre-filter the
  signal to a one-sided bandwidth of B, otherwise it will not be
  possible to accurately reconstruct the original signal from the
  samples. The frequency 2B that is the minimum sample rate to retain
  all of the signal information is called the Nyquist frequency.

  The spectrum of the sampled signal is the same as the spectrum of the
  continuous signal except that copies (known as aliases) of the
  original now appear centred on all integer multiples of the sample
  rate. As an example, if a signal of 20 kHz bandwidth is sampled at 50
  kHz then alias spectra appear from 30 - 70 kHz, 80 - 120 kHz, and so
  on. It is because the alias spectra must not overlap that a sample
  rate of greater than 2B is required.  In digital audio we are
  concerned with the base-band - that is to say the signal components
  which extend from 0 to B. Therefore, to sample at the standard digital
  audio rate of 44.1 kHz requires the input signal to be band-limited to
  the range 0 Hz to 22.05 kHz.  [Chris]

Q5.2 - What is oversampling?

  To take distortionless samples at 44.1kHz requires that the analogue
  signal be bandlimited to 22.05kHz. Since the audio band is reckoned to
  extend to 20kHz we require an analogue filter that cuts off very
  sharply between 20kHz and 22kHz to accomplish this. This is expensive,
  and suffers from all the ailments associated with analogue

  Oversampling is a technique whereby some of this filtering may be done
  (relatively cheaply and easily) in the digital domain. By sampling at
  a high rate (for example 4 times 44.1kHz, or 176.4kHz) the analogue
  filter can have a much lower slope since its transition band is now
  20kHz to 88kHz (ie half of 176kHz). The samples are then passed
  through a digital filter with a sharp cutoff at 20kHz, after which
  three of every four are discarded, resulting in the sample stream at
  44.1kHz that we require.  [Chris]

Q5.3 - What is the difference between a "1 bit" and a "multibit" converter?
       What is MASH?  What is Delta/Sigma?  Should I really care?

  Audio data is stored on CD as 16-bit words. It is the job of the
  digital to analogue converter (DAC) to convert these numbers to a
  varying voltage. Many DAC chips do this by storing electric charge in
  capacitors (like water in buckets) and selectively emptying these
  buckets to the analogue ouput, thereby adding their contents. Others
  sum the outputs of current or voltage sources, but the operating
  principles are otherwise similar.

  A multi-bit converter has sixteen buckets corresponding to the sixteen
  bits of the input word, and sized 1, 2, 4, 8 ... 32768 charge units.
  Each word (ie sample) decoded from the disc is passed directly to the
  DAC, and those buckets corresponding to 1's in the input word are
  emptied to the output.

  To perform well the bucket sizes have to be accurate to within +/-
  half a charge unit; for the larger buckets this represents a tolerance
  tighter than 0.01%, which is difficult.  Furthermore the image
  spectrum from 24kHz to 64kHz must be filtered out, requiring a
  complicated, expensive filter.

  Alternatively, by using some digital signal processing, the stream of
  16-bit words at 44.1kHz can be transformed to a stream of shorter
  words at a higher rate. The two data streams represent the same signal
  in the audio band, but the new data stream has a lot of extra noise in
  it resulting from the wordlength reduction. This extra noise is made
  to appear mostly above 20kHz through the use of noise-shaping, and the
  oversampling ensures that the first image spectrum occurs at a much
  higher frequency than in the multi-bit case.

  This new data stream is now converted to an analogue voltage by a DAC
  of short word length; subsequently, most of the noise above 20kHz can
  be filtered out by a simple analogue filter without affecting the
  audio signal.

  Typical configurations use 1-bit words at 11.3MHz (256 times over-
  sampled), and 4-bit words at 2.8MHz (64 times oversampled).  The
  former requires one bucket of arbitrary size (very simple); it is the
  basis of the Philips Bitstream range of converters. The latter
  requires four buckets of sizes 1, 2, 4 and 8 charge units, but the
  tolerance on these is relaxed to about 5%.

  MASH and other PWM systems are similar to Bitstream, but they vary the
  pulse width at the output of the digital signal processor. This can be
  likened to using a single bucket but with the provision to part fill
  it. For example, MASH allows the bucket to be filled to eleven
  different depths (this is where they get 3.5 bits from, as 2^(3.5) is
  approximately eleven).

  Lastly it is important to note that these are all simply different
  ways of performing the same function. It is easy to make a lousy CD
  player based around any of these technologies; it is rather more
  difficult to make an excellent one, regardless of the DAC technology
  employed. Each of the conversion methods has its advantages and
  disadvantages, and as ever it is the job of the engineer to balance a
  multitude of parameters to design a product that represents value for
  money to the consumer.  [Chris]

Q5.4 - On an analog recorder, I was always taught to make sure the signal
       averages around 0 VU.  But on my new DAT machine, 0 is all the way at
       the top of the scale.  What's going on here?

  Analog recorders are operated such that the signal maintains a nominal
  level that strikes a good balance between signal-to-noise ratio and
  headroom.  Further, since analog distorts very gently, you often can
  exceed your headroom in little bits and not really notice it.

  Digital is not nearly as forgiving.  Since digital represents audio as
  numerical values, higher levels will eventually force you to run out
  of numbers.  As a result, there is an absolute ceiling as to how hot
  you can record.  If you record analog and have a nominal 12 dB of
  headroom, you'll probably be okay if you have one 15 dB transient that
  lasts for 1/10th of a second.  The record amps _might_ overload, the
  tape _might_ saturate, but you'll probably be fine.  In a digital
  system, those same 3 dB of overshoot would cause you to clip hard.  It
  would not be subtle or forgiving.  You would hear a definite snap as
  you ran out of room and chopped the top of your waveform off.

  The reality is that digital has NO HEADROOM, because there is no
  margin for overshoot.  You simply must make sure that the entire
  dynamic range of the signal fits within the limits of the dynamic
  range of your recorder, without exception.  The only meaningful
  absolute on a digital recorder, therefore, is the point at which you
  will go into overload.  The result is the metering system we now have.
  0 dB represents digital ceiling, or full-scale.  The negative numbers
  on the scale represents your current level relative to the ceiling.

  Thus, to return to our example, if you have a transient with 15dB of
  overshoot past your nominal level, you must then place your nominal
  level at a maximum of -15 dB.  0 dB on the meters is the absolute limit
  of what you can record.  [Gabe]

Q5.5 - Why doesn't MiniDisc or Digital Compact Cassette sound as good as DAT
        or CD?  After all, they're both digital.

  Both MD and DCC use lossy compression algorithms (called ATRAC and
  PASC respectively); crudely, this means that the numbers coming out of
  the machine are not the same as those that went in. The algorithms use
  complex models of the way the ear works to discard the information
  that it thinks would not be heard anyway.

  For example, if a pin dropped simultaneously with a gunshot, it may be
  reasonable to suggest that it isn't worth bothering to record the
  sound of the pin! In fact it turns out that around 75 to 80 per cent
  of the data for typical music can be discarded with surprisingly
  little quality loss.

  However, nobody denies that there is a quality loss, particularly
  after a few generations of copying. This fact and others make both MD
  and DCC useful only as a consumer-delivery format. They have very
  little use in the studio as a recording or (heaven forbid!) mastering
  format.  [Chris]

Q5.6 - What is S/P-DIF?  What is AES/EBU?

  AES/EBU and S/P-DIF describe two similar protocols for communicating
  two-channel digital audio information over a serial link. They are
  slightly different in details, their basic format is almost identical,
  but there are enough differences that the two are, for all intents and
  purposes electrically incompatible.  Both of these digital protocols
  are described fully in an international standard, IEC 958, available
  from the International Electrotechnical Commission.

  AES/EBU (which stands for the joint Audio Engineering Society/European
  Broadcasting Union standard) is the so-called "professional" protocol.
  It uses standard 3-pin XLR connectors and 110-ohm balanced
  differential cables for connection (no, standard microphone cables,
  not even good quality cables, won't work, even though it seems they
  might) and a 5 volt, differential signal.

  S/P-DIF (which stands for Sony/Philips Digital InterFace, a now
  obsolete standard superseded by IEC 958) is the so-called "consumer"
  format. It uses what appears to be standard RCA connectors and cables,
  but, in fact, require 75-ohm connectors and cables. Good quality video
  "patch" cables have proven adequate (no, standard "audio" patch cords,
  even excellent quality versions, have been shown not to work). The
  signals are 0.5 volts unbalanced.

  The actual datastream, are very similar.  Each sample period a "frame"
  is transmitted. Each frame consists of two "subframes", one each for
  left and right channels, each subframe is 32 bits wide. In that
  subframe, 4 bits are used for synchronization, then up to 24 bits are
  usable for audio (the "consumer mode" format is limited to 16 bits).
  The remaining four bits are used for parity (the first level of error
  detection), validity, user status and channel status. 192 subframes
  are collected, and the 192 user bits and 192 channel status bits are
  collected into separate 24 8 bit status bytes for each channel.

  The channel status bytes are interesting, because they contain the
  important control information and the major differences between the
  two protocol formats. One bit tells whether the data stream is
  professional or consumer format.  There are bits that specify
  (optionally) the sample rate, deemphasis standards, channel usage, and
  other information.  The consumer format has several bits allocated to
  copy protection and control: the SCMS bits.

  Now, the notion that all of this is encoded in a standard may be
  reassuring, but a standard is nothing but a voluntary statement of
  common industry practice. There is a lot of incompatibility between
  equipment out there caused directly by subtle differences between
  interpretations and implementations.  The result is that some
  equipment simply refuses to talk to each other.  Even THAT 
  possibility is stated in the standard!  [Dick]

Q5.7 - What is clock jitter?

  Clock jitter is a colloquialism for what engineers would readily call
  time-domain distortion.  Clock jitter does not actually change the
  physical content of the information being transmitted, only the time
  at which it is delivered.  Depending on circumstance, this may or may
  not affect the ultimate decoded output.

  Let's look at this a little more closely.  Digital audio is sent as a
  set of binary digits....1's and 0's.  But that is only a logical
  construct.  In order to transmit binary math electrically, we use
  square waves.  Realize that although we have two mathematical states,
  we have to transmit such a construct using control voltages and

  All digital audio systems start with a crystal controlled oscillator
  producing a square wave signal that is used to synchronize the entire
  digital audio sampling and playback processes.  Now, for a clock, we
  don't really care about the fact that the clock might be at state 1 or
  state 0 at any given moment.  That doesn't give us any information.
  As a computer, I can't tell if my clock has just gotten to state 1, or
  if it's been sitting there for a microsecond.  Thus it isn't the
  states we care about.  Instead, we care about the state
  *changes*....when the clock shifts from one state to the other.

  Now, in a perfect square wave (no such thing exists), the change of
  state would be instantaneous.'s done.  But in reality, it
  doesn't work this way.  Square waves contain high orders of harmonics.  
  Fourier teaches us that all complex waveforms are made up of simpler
  waveforms.  Thus, as we run through noisy electronics, long cables,
  inadvertent filtering circuits, we begin to lose some of our
  harmonics.  When this happens, our square wave begins to lose form.

  The result of this is that our nice sharp corners become rounded.  So
  our state changes are no longer precisely at the edge anymore, because
  there is no more edge.  The pointy edge is now all fuzzy.  It now
  depends on design of the electronic comparator circuit as to when the
  clock state will change, as the stage change has shifted.  The clock
  is, essentially, jittering.

  People love to bark out "Bits is bits.  A copy of a computer file
  works as well as the original."  Yes, this is true.  But these
  jittering bits can create audible distortion during the digital-to-
  analog conversion, and the industry is working hard to reduce the
  amount of jitter present in digital systems.

  Furthermore, emerging research is suggesting that certain types of
  jitter may produce digital copies with eccentricities that result in
  more jittery output on playback.  The jury is still out on the
  specifics however.  Stay tuned.  [Gabe]

Q5.8 - What kind of cable AES/EBU or S/P-DIF cables should I use? How long
       can I run them?

  The best, quick answer is what cables you should NOT use!

  Even though AES/EBU cables look like orinary microphone cables, and S/P-DIF
  cables look like ordinary RCA interconnects, they are very different.

  Unlike microphone and audio-frequency interconnect cables, which are
  designed to handle signals in the normal audio bandwidth (let's say that
  goes as high as 50 kHz or more to be safe), the cables used for digital
  interconnects must handle a much wider bandwidth. At 44.1 kHz, the digital
  protocols are sending data at the rate of 2.8 million bits per second,
  resulting in a bandwidth (because of the biphase encoding method) 
  of 5.6 MHz.

  This is no longer audio, but falls in the realm of bandwidths used by
  video. Now, considerations such as cable impedance and termination become 
  very important, factors that have little or no effect below 50 kHz.

  The interface requirements call for the use of 110 ohm balanced cables for
  AES/EBU interconnects, and 75 ohm coaxial unbalanced interconnects for
  S/P-DIF interconnects. The used of the proper cable and the proper
  terminating connectors cannot be overemphasised. I can personally testify
  (having, in fact, looked at the interconnections between many different
  kinds of pro and consumer digital equipment) that ordinary microphone or
  RCA audio interconnects DO NOT WORK. It's not that the results sound
  subtly different, it's that much of the time, it the receiving equipment
  is simply unable to decode the resulting output, and simply shuts down.

  Fortunately, there is a ready solution for S/P-DIF cables. Any store that
  sells high quality 75 ohm RCA video interconnect (or "dubbing") connectors
  also sells high-quality S/P-DIF interconnects as well. They may not know
  it, but they do. This is because the signal and bandpass requirements for
  video and S/P-DIF cables are the same. National chains such as Radio Shack
  sell such cables, and the data seems to indicate that they are good digital

  For AES/EBU, there are fewer, less common solutions. Companies such as
  Canare make excellent cables. Professional audio suppliers and distributors
  may be good sources for such cables. If you are handy with a soldering
  iron, then you can purchase 110 ohm balanced shielded cable and make your
  own (which I have done quite successfully). Cables such as Alpha Twinax,
  Carol Twin Coaxial, Belden 9207 twin axial, and the like, all work well for
  this application. Use high-quality XLR connectors (be warned that these
  cables are 0.330 inches in diameter and are a VERY tight fit in the
  neoprene strin reliefs of many connectors: warming them in hot water makes
  them pliable enough to work well).

  As to how long these cables can be, it's hard to say. However, a couple of
  general rules apply.

  S/P-DIF was NEVER intended to be a long-haul hardware interconnect. The
  relevant specifications talk of interconnect lengths less than 10 meters
  (33 feet). In fact, many pieces of equipment cannot tolerate cables even
  that long, due to the excessive capacitance and possibly induced common
  mode interference.

  AES/EBU is more tolerant of longer runs because it is balanced (thus more
  immune to interference) and it's run at a higher signal level (5 volts
  instead of 0.5 volts). The standards "allow signal transmission up to a few
  hundred meters in length."

  The reality is that much is highly dependent upon the actual conditions at
  hand. The requirements are that the received signal fit within certain
  requirements of rise time/period and voltage level, the so-called "eye
  diagram". In other words, regardless of what kind of cable you use, if it
  can't move the voltage at the receiver far enough soon enough, it simply
  isn't going to work.

  Another complicating factor is that both protocols allow a degree of
  multi-drop capability. This means a single transmitter can drive several
  receivers (the last of which must be terminated with the proper termination
  impedance). However, implementing multi-drop puts more stingent
  requirements on impedance matching. [Dick]

Q5.9 - What is SCMS?  How do I defeat it?

  SCMS is the Serial Copy Management System, a form of copy protection
  that was mandated by Federal law (the Home Recording Rights Act).
  SCMS consists of a set of subcode flags that indicate to a digital
  recorder whether or not the source may be copied.  Under the HRRA,
  consumers are permitted to make one digital generation, but no more.
  Thus when, for instance, the consumer copies a CD onto DAT, the SCMS
  flag is set on the copy, and no further generations can be made.

  SCMS is only mandated in consumer machines.  Any recorder sold 
  through professional channels, and which is intended for use in
  professional applications, does not have to implement it.

  There are several professional products, such as Digital Domain's
  FCN-1 format converter, which allow manipulation of the SCMS flags.
  These units exist so that professional engineers may adjust the 
  subcode bits of the recordings they produce.  [Gabe]

Q5.10 - What is PCM-F1 format?

  In the 1980s, before the DAT era, Sony produced a set of PCM adaptors
  that enabled one to record digital audio using a video cassette
  machine.  These units had RCA audio connections for input and output,
  as well as video I/O that could be sent to, and received from, the
  VCR.  At the time, these systems offered performance far in excess of
  conventional analog recorders available in the price category.

  Sony released many models, including the PCM-F1, PCM-501, PCM-601, and
  PCM-701.  Perhaps the most interesting is the PCM-601, which has
  S/P-DIF digital I/O.  These units are highly prized since they are the
  only units that can be used to make digital transfers of F1 tapes to
  modern hardware.

  There are some engineers who insist that, despite the clunkiness of the
  format by modern DAT standards, the F1 series was the best digital format
  ever developed.  To this day, it is not surprising to see an F1 encoder 
  on a classical recording session. [Gabe]

Q5.11 - How do digital recorders handle selective synchronization?

  Selective Synchronization, or "sel-sync" as it is often called, is the
  ability of a recorder to play and record simultaneously, allowing
  synchronous recording of new material onto specific tracks without 
  erasing everything on tape.  This technique is what makes overdubbing 

  On an analog recorder, audio tracks are discrete entities, and the
  sync head is really just a stack of individual heads, any one of which
  is capable of recording or playing back.  Thus sel-sync is a
  relatively simple matter of putting some heads into record and others
  into repro.

  In the digital world, the problem is highly complex.  First, A/D and
  D/A conversion involves an acquisition delay of several milliseconds.
  Second, and more importantly, digital tracks are not discrete.  Rather,
  they are multiplexed together on a tape, along with subcode and other
  non-audio information.  So how can you replace one track and leave the
  others untouched?

  The answer is a technique called "read before write" (RBW) or "read,
  modify, write" (RMW) which involves a second set of heads.  The data
  is read from the tape and flushed into a buffer, where it can be
  modified, and ultimately written back to the tape.  Thus when you
  "punch in" on a digital deck, you are physically re-writing all the
  tracks, not just the one you're overdubbing.  You are not, however,
  changing the data on any track other than the one you want to 
  replace.  [Gabe]

Q5.12 - How can a 44.1 kHz sampling rate be enough to record all the 
harmonics of music?  Doesn't that mean that we chop off all the harmonics
above 20 khz?  Doesn't this affect the music?  After all, analog systems
don't filter out all the information above 20 kHz, do they?

  This whole question is based on the premise that "analog systems don't
  filter out all the information above 20 kHz." Indeed there are mixers
  and power amplifiers and other electronic systems that are capable of
  stunningly wide bandwidth, often exceeding 100 kHz, the same cannot be
  said for the entire analog reproduction chain. The mechanical
  transducers, microphones, speaker and phono cartridges seldom have
  real response far exceeding 20 kHz. In fact, some of the most highly
  regarded large diaphragm condensor microphones often used in very high
  quality recordings seldom exceed 18 kHz bandwidth. Analog tape
  recorders rarely have bandwidths as wide as 25 kHz, and LP
  reproduction systems have similar limitations in reality.

  So while it may be possible to send very high frequency ultrasonic
  signals through parts of both analog and digital reproduction chains,
  there are, in both technologies, fundamental and insurmountable limits
  to the bandwidth that, in reality, lead to very similar actual
  reproducible bandwidths in each.

  Thus, one of the basic premises of the question is flawed. Analog
  systems DO filter out information above 20 kHz. Further, the frequency
  response and phase errors of even the very best well-maintained analog
  reproduction systems have response errors far exceeding those of even
  middle of the line digital equipment. Whether one person may find
  those errors tolerable or even likeable or not is a matter or personal
  preference that is beyond the scope of this or any other technical

  There are a variety of anecdotal tales that are advanced to "prove"
  that the ear can hear far beyond what is conventionally accepted as
  the 20 kHz upper limit (an upper limit that, for the most part,
  applies to young people only: modern high SPL music and noise levels
  has lead to a widespread deterioration in the hearing of the adult
  population at large, and especially amongst young males).

  For example, there is an apocryphal story about Rupert Neve that
  tells of a console channel that sounded particularly "bad". It was
  later discovered that it was oscillating at some ultrasonic frequency,
  like 48 kHz. Rupert Neve is rumored to have seized upon this as
  "proof" that the ear can hear well beyond 20 kHz. However, there exist
  an entire range of perfectly plausible mechanisms that require NO
  ultrasonic acuity to detect such a problem. For example, the existence
  of ANY nonlinearity in the system would result in the production of
  intermodulation tones that would fall well within the 20 kHz audio
  band and certainly would make it sound awful. Even the problem that
  was causing the oscillation itself could lead to massive artifacts at
  much lower frequencies that would completely account for the alleged
  sound of the mixer in the complete absence of a 48 kHz "whistle."

  Whether 20 kHz is an adequate bandwidth is a debatable subject.
  However, several important facts have to be remembered. First, BOTH
  analog AND digital reproduction systems suffer from roughly the same
  bandwidth limiting. Second, digital systems using properly implemented
  oversampling techniques have far less severe phase and frequency
  response errors within the audible band. No analog storage and
  reproduction system can match the phase and response linearity of a
  digital system, both at low and high frequencies. Once those
  demonstrable facts are acknowledged, then the discussion about
  supra-20 kHz aural detectability can continue, knowing that, if it is
  demonstrated to be significant, both systems are provably deficient.

Q5.13 - Yeah, well what about square waves? I've seen square wave
tests of digital systems that show a lot of ringing. Isn't that bad?

  Square waves are a mathematically precisely defined signal. One of the
  ways to describe a perfect square wave is as the sum an infinite series
  of sine waves in a precise phase, harmonic and amplitude relationship.
  The relation is:

                     1           1           1           1
    F(t) = sin(wt) + -sin(3wt) + -sin(5wt) + -sin(7wt) + -sin(9wt) ...
                     3           5           7           9

  where t is time, w is "radian frequency", or 2 pi times frequency.
  Remember, we require an infinite number of terms to describe a perfect
  square wave. If we limit the number of terms to, say, 10 terms, (such as
  the case with a 1 kHz square wave perfectly band limited to 20 kHz),
  there simply aren't enough terms to describe a perfect square wave.
  What will result is a square wave with the highest harmonic imposed on
  top as "ringing." In fact, this appearance indicates that the phase
  and frequency response is perfect out to 20 kHz, and the bandwidth
  limiting is limiting the number of terms in the series.

  Well, what would a perfect analog system do with square waves? As it
  turns out, if you take a high quality 15 IPS tape recorder, bias and
  adjust it for the flattest possible frequency response over the widest
  possible bandwidth, the result looks remarkably like that of a good
  digital system for exactly the same reasons. 
  On the other hand, adjust the analog tape recorder for a square wave
  response that has no ringing, but the fastest possible rise time. Now
  listen to it: it sounds remarkably dull and muffled compared to the
  input. Why? Because in order to achieve that square wave response, it's
  necessary to severely roll off the high end response in order to
  suppress the high-frequency components needed to achieve fastest rise
  time. [Dick]

Q5.14 - How can a 16-bit word length be enough to record all the detail
in music?  Doesn't that mean that the sound below -96 dB gets lost in the
noise? Since it is commonly understood that humans can perceive audio
that IS below the noise floor, aren't we losing something in digital
that we don't lose in analog?

  You're correct in saying that human hearing is capable of perceiving
  audio that is well below the noise floor (we won't say what kind of
  noise floor just yet). The reason it can do this is through a process
  the ear and brain employ called averaging.
  If we look at a single sample in a digital system or an instantaneous
  shapshot in an analog system, the resulting value that we measure will
  consist of some part signal and some part ambiguity. Regardless of the
  real value of the signal, the presence of noise in the analog system
  or quantization in the digital system sets a limit on the accuracy to
  which we can unambiguously know what the original signal value was. So
  on an individual sample or instantaneous snapshot, there is no way
  that either ear or measurement instrument can detect signals that are
  buried below either the noise or the quantization level (when properly
  However, if we look at (or listen to) much more than a single sample,
  through the process of averaging, both instruments and the ear are
  capable of detecting real signals below the noise floor. Let's look at
  the simple case of a constant voltage that is 1/10th the value of the
  noise floor. At the instantaneous or sample point, the noise value
  overwhelms the signal completely. But, as we collect more consecutive
  snapshots or samples, an interesting thing begins to happen. The noise
  (or dither) is random and its long term average is, in fact, 0. But the
  signal has a definite value, 1/10. Average the signal long enough, and the
  average value due to the noise approaches 0, but the average value of
  the signal remains constant at 1/10.
  A somewhat analogous process happens with high frequency tones. In
  this case the averaging effect is that of a narrow-band filter. The
  spectrum of the noise (or simple dither) is broadband, but the
  spectrum of the tone is very narrow band. Place a filter centered on
  the tone and while we make the filter narrower and narrower, the
  contribution of the noise gets less and less, but the contribution of
  the signal remains the same.
  Both the ear and measurement instruments are capable of averaging
  and filtering, and together are capable of pulling real signals from
  deep down within the noise, as long as the signals have one of two
  properties: either a period that is long compared to the inherent
  sampling period of the signal in a digital system or long compared to
  the reciprocal of the bandwidth in an analog system, or a periodic
  signal that remains periodic for a comparably long time.
  Special measurement instrument were developed decades ago that were
  capable of easily detecting real signals that were 60 dB below the
  broadband noise floor. And these devices are equally capable of
  detecting signals under similar conditions in properly dithered
  digital systems as well.

  How much the ear is capable of detecting is dependent upon many
  conditions, such as the frequency and relative strength of the tone,
  as well as individual factors such as aging, hearing damage and the

  But the same rules apply to both analog systems with noise and digital
  systems with decorrelated quantization noise. [Dick]

Q5.15 -  Q5.14 - What's all this about 20- and 24-bit digital audio?  Aren't
         CDs limited to 16 bits?

  Yes, CDs are limited to 16 bits, but we can use >16-bit systems to produce
  16-bit CDs with higher quality than we could otherwise.

  We are able to record audio with effective 20-bit resolution nowadays.
  The finest A/D converter systems have THD+N values around -118 dB with
  linearity extending far below even that.  When it comes time to reduce
  our word-length to 16 bits, we can use any one of a variety of noise
  shaping curves, the job of which is to mix with our 24-bit audio, shift
  the dither spectrum of the noise into areas where our ears are less
  sensitive, thus enabling the noise component to comprise audio information
  at the spectral areas where our ears are most sensitive.  See Lipschitz's
  seminal papers for fuller detail on this subject.

  Furthermore, we often perform DSP calculations on our audio, and to that
  end it is worthwhile to carry out the arithmetic with as much precision
  as we can in order to avoid rounding errors.  Most digital mixers carry
  their math out to 24-bit precision at the I/O, with significantly longer
  word lengths internally.  As a result, two 16-bit signals mixed together
  can produce a valid 24-bit output word.  For that matter, a 16-bit signal
  subjected to a level change can produce a 24-bit output if desired (except,
  of course, for a level change that is a multiple of 6 dB, as that's just
  a shift left or right).

  The number of noise shaping curves available today is staggering.  Sony
  SBM, Weiss, Meridian 618, Sonic TBM, Apogee UV-22, Prism SNS, Lexicon 
  PONS, Waves, and, of course, the classic Lipschitz curve are just a few
  of the multitudinous options that now exist. [Gabe]

Section VI - Digital editing and mastering

Q6.1 - What is a digital audio workstation?

  A digital audio workstation (DAW) is one of our newest audio buzzwords,
  and applies to nearly any computer system that is meant to handle or 
  process digital audio in some way.  For the most part however, the
  term refers to computer-based nonlinear editing systems.  These systems
  can comprise a $500 board that gets thrown into a PC, or can refer to
  a $150,000 dedicated digital mastering desk.  [Gabe]

Q6.2 - How is digital editing different from analog editing?

  In the days of analog editing, one edited with a razor blade and a
  diagonal splicing block.  Making a cut meant scrubbing the tape over
  the head, marking it with a grease pencil, cutting, and then taping 
  the whole thing back together.  Analog editing (particularly on music)
  was as much art as it was craft, and good music editors were worth
  their weight in gold.

  In many circles, analog editing has gone the way of the Edsel,
  replaced by digital workstation editing.  For complex tasks, DAW-based
  editing offers remarkable speed, the ability to tweak an edit after
  you make it, a plethora of crossfade parameters that can be optimized
  for the edit being made, and most importantly, the ability to undo
  mistakes with a keystroke.  Nearly all commercial releases are being
  edited digitally nowadays.  Since satisfactory editing systems can
  be had for around $1,000, even home recordists are catching onto the
  advantages.  More elaborate systems can cost tens of thousands of

  There are certain areas where analog editing still predominates,
  however.  Radio is sometimes cited as an example, though this has begun
  to change thanks to products like the Orban DSE 7000.  The needs of 
  radio production are often quite different from those of music editors,
  and a number of products (the Orban being a fine example) have sprung
  up to fill the niche.  Nonetheless, in spite of the rapid growth of
  DAWs in the radio market, razor blades are still found in daily use
  in radio stations.  [Gabe]

Q6.3 - What is mastering? 

  Mastering is a multifaceted term that is often misunderstood.  Back in
  the days of vinyl records, mastering involved the actual cutting of
  the master that would be used for pressing.  This often involved a
  variety of sonic adjustments so that the mixed tape would ultimately
  be properly rendered on vinyl.

  The age of the CD has changed the meaning of the term quite a bit.
  There are now two elements often called mastering.  The first is the
  eminently straightforward process of preparing a master for pressing.
  As most mixdowns now occur on DAT, this often involves the relatively
  simple tasks of generating the PQ subcode necessary for CD replication.
  PQ subcode is the data stream that contains information such as the 
  number of tracks on a disc, the location of the start points of each
  track, the clock display information, and the like.  This information
  is created during mastering and prepared as a PQ data burst which the
  pressing plant uses to make the glass pressing master.

  Mastering's more common meaning, however, is the art of making a
  recording sound "commercial."  Is is the last chance one has to get
  the recording sounding the way it ought to.  Tasks often done in
  mastering include: adjustment of time between pieces, quality of
  fade-in/out, relation of levels between tracks (such that the listener
  doesn't have to go swinging the volume control all over the place),
  program EQ to achieve a desired consistency, compression to make one's
  disc sound LOUDER than others on the market, the list goes on.

  A good mastering engineer can often take a poorly-produced recording
  and make it suitable for the market.  A bad one can make a good
  recording sound terrible.  Some recordings are so well produced,
  mixed, and edited that all they need is to be given PQ subcode and
  sent right out.  Other recordings are made by people on ego trips, who
  think they know everything about recording, and who make recordings
  that are, technically speaking, wretched trash.

  Good mastering professionals are acquainted with many styles of music,
  and know what it is that their clients hope to achieve.  They then use
  their tools either lightly or severely to accomplish all the multiple
  steps involved in preparing a disc for pressing. [Gabe]

Q6.4 - What is normalizing?

  Normalizing means bringing a digital audio signal up in level such
  that the highest peak in the recording is at full scale.  As we saw in
  Q5.4, 0 dB represents the highest level that our digital system can
  produce.  If our highest level is, for instance, -6 dB, then the
  absolute signal level produced by the player will be 6 dB lower than
  it could have been.  Normalizing just maximizes the output so that the
  signal appears louder.

  Contrary to many frequently-held opinions, normalizing does NOT
  improve the dynamic range of the recording in any way, since as you
  bring up the signal, you also bring up the noise.  The signal-to-noise
  ratio is a function of the original recording level.  If you have a
  peak at -6 dB, that's 6 dB of dynamic range you didn't use, and when
  you normalize it to 0 dB, your noise floor will rise an equivalent

  Normalizing may help optimize the gain structure on playback, however.
  Since the resultant signal will be hotter, you'll hear less noise from
  your playback system.

  But the most common reason for normalizing is to make one's recordings
  sound, LOUDER, BRIGHTER, and have more PUNCH, since we all know that
  louder recordings are better, right? :-)  [Gabe]

 Q6.5 - I have a fully edited DAT that sounds just like I want it to sound on
        the CD.  Is it okay to send it to the factory?

  This is a highly case-specific question.  Some people truly have the
  experience to produce DATs on mixdown or editing that are ready to go
  in every conceivable way.  Often these people can send their tapes out
  for pressing without the added expense of a mastering house.

  However, if you do not have this sort of expertise, and if your only
  reason for wanting to send it out immediately is because you know that
  it is technically possible to press from what you have, then you would
  be advised to let a mastering engineer listen to and work on your
  material.  A good mastering engineer will often turn up problems and
  debatable issues that you didn't even know were there.  Also, any 
  decent mastering house will provide you with a master on a format 
  significantly more robust than DAT.  

  DAT is a fine reference tape format, but it simply is not the sort of
  thing you want to be sending to a pressing plant. [Gabe]

Q6.6 - What is PCM-1630?  What is PMCD?

  PCM-1630 is a modulation format designed to be recorded to 3/4"
  videotape.  It was, for many years, the only way one could deliver a
  digital program and the ancillary PQ information to the factory for
  pressing.  The PCM-1630 format is still widely used for CD production.

  But PCM-1630 is now certainly an obsolete system, as there are many 
  new formats that are superior to it in every way.  One of the most
  popular formats for pressing now is PMCD (Pre-Master Compact Disc).
  This format, developed by Sonic Solutions, allows for CD pressing 
  masters to be written out to CD-Rs that can be sent to the factory
  directly.  These CD-Rs contain a PQ burst written into the leadout
  of the discs.

  Some plants have gone a step further and now accept regular CD-Rs,
  Exabyte tapes, or even DATs for pressing.  The danger here is that
  some users may think that they can prepare their own masters without
  the slightest understanding of what the technical specifications are.
  For instance, users preparing their own CD-Rs must do so in one
  complete pass.  It is not permitted, for instance, to stop the CD-R
  deck between songs, as this creates unreadable frames that will cause
  the disc to be rejected at the plant.  [Gabe]

Q6.7 - When preparing a tape for CD, how hot should the levels be?

  Ideally you should record a digital master such that your highest
  level is at 0 dB, if only to maximize the dynamic range of your
  recordings.  Many people like their CDs to be loud, and thus they
  will normalize anyway even if they don't hit 0 dB during recording.

  Some classical recordings are deliberately recorded with peaks 
  that are significantly lower than 0 dB.  This is done in order to
  prevent quiet instruments such as lutes and harpsichords from
  being played too loud.  If you record a quiet instrument such as
  a harpsichord out to 0 dB, the listener would have to put the 
  volume control all the way at the bottom in order to get a
  realistic level, and the inclination would be to play it at a
  "normal" listening level.  By dropping the mastering level, the
  listener is more likely to set the playback level appropriately.

Q6.8 - Where can I get CDs manufactured?

[To come]

Q6.9  - How are CD error rates measured, and what do they mean?

[Forthcoming.  -Gabe]

Section VII - Market survey.  What are my options if I want --

Q7.1 - A portable DAT machine

Q7.2 - A rack size DAT machine

Q7.3 - An inexpensive stereo microphone

Q7.4 - An inexpensive pair of microphones for stereo

Q7.5 - A good microphone for recording vocals

Q7.6 - A good microphone for recording [insert instrument here]

Q7.7 - A a small mixer

Q7.8 - A portable cassette machine

Q7.9 - A computer sound card for my IBM PC or Mac

Q7.10 - An eight-track digital recorder?

Section VIII - Sound reinforcement

 Q8.1 - We have a fine church choir, but the congregation can't hear them.
        How do we mic the choir?

 Q8.2 - How do I 'ring out' a system?

 Q8.3 - How much power to I need for [insert venue here]?

 Q8.4 - How good is the Sabine feedback eliminator?

Section IX - Sound restoration

Q9.1 - How can I play old 78s?  

  First rule of thumb:  DO NOT PLAY THEM WITH AN LP STYLUS!  

  The grooves on 78s are gigantic compared to the microgroove LPs.  
  As a result, specialized styli are needed for proper playback of 78s.
  Also, the RIAA equalization curve normally used for LPs has no
  relation to the frequencies that were equalized on 78 recordings.

  The easiest stylus to obtain is the Shure V15 with the 78 pickup.  This
  is a good, though not great, 78 stylus and will do an okay job at playing
  most discs. 

  The serious 78 collector will want to obtain not only a collection of
  78 styli for the various discs in their collection (groove sizes varied,
  and most serious collectors own a handful of styli), but also a preamp
  with variable equalization curves.  One supplier of all this apparatus
  is Audio-78 Archival Supplies at (415) 457-7878.  [Gabe]

Q9.2 - How can I play Edison cylinders?

  Edison cylinders are best played back with Audio 78's adaptor.  You
  remove the horn and reproducer element from your cylinder machine,
  install their electric reproducer, and connect it up to a phono
  preamp, sans RIAA. [Gabe]

Q9.3 - What are "Hill and Dale" recordings, and how do I play them back?

  Hill & Dale recordings are discs where the grooves move vertically 
  instead of horizontally.  Edison, for instance, cut his discs this 
  way.  In order to play Edison discs, one needs a special glass ball
  stylus that is 3.7-4.0 mil wide.  This is available from Audio 78,
  among other places.

  Also, since the information is vertical instead of horizontal, one
  must rewire a stereo phono cartridge to reject the normal horizontal
  information and reproduce only the normally-discarded vertical
  information.  This is easily accomplished by wiring a stereo cartridge
  in mono, summing the channels, with one channel out of phase.  In
  other words, connect the cartridge as follows:

			  |         ____  to preamp
		+ L -     + R -    |
		|   |         |    |
		|   |_________|    |
  One caveat: if your cartridge has one lug shorted to ground, make
  sure that this lug is connected to the ground on your preamplifier.
  It doesn't actually matter which channel you invert.

  Some preamps like the FM Acoustics 222 or the OWL 1 have a switch
  that will do this for you without rewiring.  [Gabe]

Q9.4 - What exactly are NoNOISE and CEDAR?  How are they used?

  NoNOISE and CEDAR are systems for noise removal.  Both of them approach
  the same sorts of noise, but use different algorithms and have different
  user interfaces, often with differing effectiveness.

  Noise can be broken down into several categories:

	IMPULSIVE NOISE:  Pops, clicks, thumps, snaps.
	CRACKLE:  The low-level "bacon frying" effect heard on 
		  LPs and 78s.
	HISS:     Tape hiss, surface noise, amplifier hiss, broadband noise
	BUZZ:	  60 Hz hum, any other steady-state noise that is relatively

  NoNOISE and CEDAR are two (expensive) techniques for removing many of
  these ailments.  It is rare that it is possible to remove all of the
  problem, nor is it ever possible to remove it with no degradation of
  the program material.  [Gabe]

 Q9.5 - How do noise suppression systems like NoNOISE and CEDAR work?

Digital techniques have been applied to many facets of sound processing
and recording and have, on the whole, been found to give results far
superior to their analogue counterparts. Nowhere is this more true than
in the field of audio restoration, where excellent processes have been
developed for removal of impulsive noise (thumps, clicks and ticks) and
attenuation of continuous broadband noise (such as tape hiss). Example
techniques for these two are outlined below.

   Impulsive noise

In this category we include many types of disturbance, from the click
generated by a scratch on a 78rpm disc, to the tiny tick created by a
single corrupt bit in a digital data stream. Also included is crackly
surface noise from 78's (that sounds like a frying pan), though this
requires somewhat different treatment; however the outline presented
below is fairly similar for both processes. Typically, audible clicks
are of a few microseconds to a few milliseconds in duration, and their
density can be up to a few thousand clicks per second on poor-quality

First the audio is split into short blocks of maybe 10ms duration. A
model is fitted to each block; this model can be thought of as a
description of the signal in simple mathematical terms. The model is
chosen such that musical data is a good fit, but the impulsive noise
is a poor fit.

For example, a simple model could be a sum of sinewaves, whose number,
frequencies, amplitudes and phases are the model parameters. The para-
meters are calculated such that when these sinewaves are added together
they match the musical parts of the signal accurately, but match the
impulsive noise badly.

Now the model can be thought of as a prediction of the music. In
undamaged sections the prediction is close (since music is known to
consist of a sum of sinewaves, at least approximately); during clicks
and pops etc. the prediction is poor, because the model has been
designed to match the music, and not the noise.

Now we can achieve impulsive noise removal by replacing the data that
fits the model badly (ie the clicks) with data predicted by the model,
which is known to be a close approximation to the music.

   Broadband Noise

Broadband noise is usually better tackled in the frequency domain. What
this entails is taking a block of data (as in the impulsive noise case)
but then calculating its spectrum. From the spectrum an estimate can
be made of which frequencies contain mostly signal, and which contain
mostly noise.

To help in making this discrimination we first take a "fingerprint" of
the noise from an unrecorded section, such as the lead-in groove of a
record, or a silence between movements of a symphony. This spectrum of
this fingerprint is then compared with the spectrum of each block of
musical data in order to decide what is noise and what is music.

The denoising process itself can be thought of as an automagically-
controlled, cut-only graphic equaliser. For each block, the algorithm
adjusts the attenuation of each frequency band so as to let the music
through, but not the noise. If the SNR in a particular band is high
(ie lots of signal, little noise) then the gain is left close to
unity. If the SNR is poor in a given band, then that band is heavily
attenuated.  [Chris]

 Q9.6 - What is forensic audio?  

  Forensic audio is audio services for legal applications.  Forensics 
  breaks down into four main categories.

	TAPE ENHANCEMENT:  Digital and analog processing to restore
	verbal clarity and make tapes easier to understand in a courtroom

	AUTHENTICITY:  Electronic and physical microscopic examination
	of a tape to prove that it has not been tampered with, altered,
	or otherwise changed from its original state.  Another common
	authenticity challenge is to determine whether a given was tape
	was indeed made on a given machine.

	VOICE IDENTIFICATION:  Voice ID, or voiceprinting, is the science
	that attempts to determine what was said, and by whom.  A variety
	of analog and digital analysis processes are used to analyze the
   	frequency and amplitude characteristics of a human voice and
	compare it against known samples.

	[More to come on this]	[Gabe]

Section X - Recording technique, Speakers, Acoustics, Sound

 Q10.1 - What are the various stereo microphone techniques? 

[I'm working on it!  -Gabe]

 Q10.2 - How do I know which technique to use in a given circumstance?

[This one too! -Gabe]

 Q10.3 - How do I soundproof a room?

Despite what you may have seen in the movies or elsewhere, egg crates
on the wall don't work!

First, understand what's meant by "soundproofing". Here we mean the
means and methods to prevent sound from the outside getting in, or
sound from the inside getting out. The acoustics within the room are
another matter altogether.

There are three very important requirements for soundproofing: mass,
absorption, and isolation.  Actually, there are also three others:
mass, absorption, and isolation. And to finish the job, you should
also use: mass, absorption, and isolation.

Sound is the mechanical vibration propagating through a material.  The
level of the sound is directly related to the size of those
vibrations. The more massive an object is, the harder it is to move
and the smaller the amplitude of the vibration set up in it under the
influence of an external sound. That's why well-isolated rooms are
very massive rooms.  A solid concrete wall will transmit much less
sound then a standard wood-framed, gypsum board wall. And a thicker
concrete wall transmits less than a thinner one: not so much because
of the distance, but mostly because it's heavier.

Secondly, sound won't be transmitted between two objects unless it's
mechanically coupled.  Air is not the best coupling mechanism. But
solid objects usually are. That's why well isolated rooms are often
set on springs and rubber isolators.  It's also why you may see
rooms-within rooms: The inner room is isolated from the outer, and
there may be a layer of absorptive material in the space between the
two. That's also why you'll also see two sets of doors into a
recording studio: so the sound does not couple directly through the
door (and those doors are also very heavy!).

If you are trying to isolate the sound in one room from an adjoining
room, one way is to build a second wall, not attached to the first.
This can go a long way to increasing the mechanical isolation. Try
using two sheets of drywall instead of one on each wall, and use 5/8"
drywall instead of 3/8", it's heavier.

But remember: make it heavy, and isolate it. Absorptive materials like
foam wedges or Sonex and such can only control the acoustics in the
room: they will do nothing to prevent sound from getting in or out to
begin with.  [Dick]

 Q10.4 - What is a near-field monitor?

A near field monitor is one that is design to be listened to in the
near field. Simple, eh?

The "near field" of a loudspeaker is area where the direct,
unreflected sound from the speaker dominates significantly over the
indirect and reflected sound, sound bouncing off walls, floors,
ceilings, the console. Monitoring in the near field can be useful
because the influence of the room on the sound is minimized.

Near field monitors have to be physically rather small, because you
essentially need a small relative sound source to listen to (imagine
sitting two feet away from an 18" woofer and a large multi- cellular
horn!). The physics of loudspeakers puts severe constraints on the
efficiency, power capabilities and low frequency response of small
boxes, so these small, near-field monitors can be inefficient and not
have the lowest octave of bass and not play ungodly loud. [Dick]

 Q10.5 - What are the differences between "studio monitors" and home

It depends upon who you ask.  There are speakers called "monitor"
speakers that are found almost exclusively in homes and never in

The purpose of a monitor speaker is to monitor the recording and
editing process.  If you buy the concept that they are but one tool in
the process (and probably the most frequently used single tool at
that), and if you buy the concept that your tools should be flawless,
than the requirements for a monitor speaker are easy to state (but
hard to achieve): they should be the most neutral, revealing and
unbiased possible.  They are the final link between your work and your
ears, and if they hide something, you'll never hear it.  If they color
something, you might be tempted to uncolor it incorrectly the other

There is another camp that suggests that monitor speakers should
represent the lowest common denominator in the target audience. The
editing and mix process should be done so that the results sound good
over the worst car speaker or boom box around. While such an idea has
validity as a means of verifying that the mix will sound good over
such speakers, using them exclusively for the process invites (as has
been thoroughly demonstrated in many examples of absolutely terrible
sounding albums) the possibility of making gross mistakes that simply
can't be heard in the mixing process. [Dick]

Q10.6 - My near field monitors are affecting the colors on my video
 	monitor. What can I do to shield the speakers?

  Despite a lot of folk lore and some very impressive sounding wisdom
  here on the net and in showrooms, there is effectively nothing that
  you can do to the speakers or the monitor, short of moving them away
  from one another, that will solve this problem.

  The problem comes from the magnetic field created by and surrounding
  the magnets in the loudspeaker. It's possible to design a magnet that
  has very little external field, but it can be an expensive proposition
  for a manufacturer.  If the magnets do have large external fields, the
  only technique that works is by solving the problem at the source: the
  magnet.  Special cancelling magnets are used, sometimes in conjunction
  with a "cup" shield directly around the magnet.

  You'll hear suggestions from people about placing a sheet of iron or
  steel between the speakers and the monitor. That might change the
  field, but it will not eliminate it. As often as not, it will make it

  You'll also here from people about shielding the speaker by lining the
  enclosure with lead or copper.  This method is absolutely guaranteed
  to fail: lead, copper, aluminum, tin, zinc and other such materials
  have NO magnetic properties at all, they will simply make the speaker
  heavier and won't solve the problem at all. There is but one material
  that has a shot at working: something called mu-metal, a heavy, very
  expensive, material designed for magnetic shield that requires
  extremely sophisticated and difficult fabrication and annealing
  techniques. Its cost is far greater than buying a new set of speakers
  that does not have the problem, and it may not even work if the
  majority of the offending field is radiated from the front of the
  speaker, which you obviously can't shield.

  Try moving the speakers relative to your monitor. Often, moving them
  an inch or two is enough to cure the problem or at least make it
  acceptable. Sometimes, placing the speakers on their sides with the
  woofers (the major offenders in most cases) farthest away from the
  monitor works.  [Dick]

Section XI - Industry information

Q11.1 - Is there a directory of industry resources?
Q11.2 - What are the industry periodicals?
Q11.3 - What are the industry trade organizations?
Q11.4 - Are there any conventions or trade shows that deal specifically 
        with professional audio?

Section XII - Miscellaneous

Q12.1 - How do I modify Radio Shack PZMs?


Q12.2 - Can I produce good demos at home?

 [Who wants to do this?]

 [That's a request for a writer, NOT a professional opinion!]

Q12.3 - How do I remove vocals from a song?

  You probably want a device called the Thompson Vocal Eliminator made
  by LT Sound in Atlanta, Georgia.  The device will cancel out any
  vocal that is mono and panned dead-center.  The unit works by
  filtering out the low frequencies, phase cancelling the rest of the
  signal, and then mixing the filtered bass back in.  The result is a
  signal with the center, common information cancelled out.  Sometimes
  it works well, other times it sounds awful. [Gabe]

Section XIII - Bibliography

 Q13.1 - Fundamentals of Audio Technology
 Q13.2 - Studio recording techniques
 Q13.3 - Live recording techniques
Q13.4 - Digital audio theory and practice

  * Ken C. Pohlmann, "Principles of Digital Audio," SAMS/Prentice Hall, 1993
    [Excellent introduction and explanation of all aspects of digital audio
    principals and practices]

   * John Watkinson, "The Art of Digital Audio," Focal Press, 1989

  * Francis Rumsey and John Watkinson, "The Digital Interface Handbook,"
    Focal Press, 1993
    [deals with interfacing standards and protocols for both audio and video]

  * IEC Standard Publication 958, "Digital Audio Interface," International
    Electro-Technical Commission, 1989, 1993
    [THE standard!]

  * Claude E. Shannon and Warren Weaver, "The Mathematical Theory of
    Communication," U. Chicago Press, 1963.  [The seminal work on digital
    sampling...includes the original 1948 BSTJ sampling paper]

Q13.5 - Acoustics

  * Leo Beranek, "Acoustics," New York, American Institue of Physics, 1986
    [The bible, but heavily mathematical, very thick and obtuse, a good book
     despite the Lincoln Hall disaster.]

  * F. Alton Everest, "The Master Handbook of Acoustics," Tab Books 1989
    [A good entry level text on acoustics, studio and listening room design,
     not heavily mathematical].

  * Arthur H. Benade, "Fundamentals of Musical Acoustics," New York, Dover
    Publications, 1990
    [A thorough book, more on the acoustics of PRODUCING music rather than

  * Any good physics text book is useful for keeping the bunk at bay.

Q13.6 - Practical recording guides

Section XIV - Miscellaneous

Q14.1 - Who wrote the FAQ?

  [Mini-bios of the FAQ day]
Q14.2 - How do you spell and pronounce the FAQ maintainer's surname?

  Those who paid attention during fourth grade English learned the rule 
  "I before E except after C."  There is no C in my name, and therefore
  the I comes before the E.  My last name is spelled "Wiener" not "Weiner."

  And it rhymes with cleaner, and starts with a W, not a V.  [Gabe]


Gabe Wiener   Dir., PGM Early Music Recordings |"I am terrified at the thought 
A Div. of Quintessential Sound, Inc., New York | that so much hideous and bad
Recording-Mastering-Restoration (212) 586-4200 | music may be put on records       | forever."--Sir Arthur Sullivan