Findings

- 8 K5k "clicks"(DHL, DHE levels, DHL Total Gain) seem to be roughly equivalent to 6dB.
- The DFL is not transposed with DCO pitch. (It follows pitch bend and pitch modulation though.)
- The DCO waves and the DFL LFO are key synced.
The DFL is a LOW pass filter. Outside of its range the level is zero.  Either I must have been wrong or this is changed in K5kOS 4.
- Resynthesis. The K5k seems to be "physically correct". Analysing a K5k sound and then reconstructing
it using 0-phase results in a similar looking and sounding waveform.

1. Analysing a single cycle of an acoustic sound and resynthesizing it using 0-phase often results in a  slightly less pleasant sound than the original (less smooth sounding). One can hear some resemblance though.

            Original(Ac.Guitar)                   Resynthesized by Igor

                                                        Analysis result



2. Analysing of the whole sound is performed by finding,  isolating and executing a FFT on each cycle of the wave.
Phase information information is ignored.
These spectrum slices are then put together to form the time varying spectrum (see below).
Time in seconds flows upwards in the 'flat' images.
Colour is like on a map (see the 3d version to the right) where blue is sea level (low).
Every vertical strip is the (DHL 0-127) level over time. The horizontal scale could have been 1-64.

Portion of original AC. Guit. sample

Portion of the resynthesized wave (by Igor using the first 64 harmonics)

So, how was the result? The guitar above was ok, but most often the result is disappointing. It's not enough that one thinks:
-'Hey, that's a piano'. It must be a piano one would want to use.


<<--Go back!