Consider a pulse train with a pulse starting time n=0, with pulses
repeated at a rate of Wo, the fundamental frequency. A pulse train
- in the time domain is equivalent to a pulse train in the frequency
+ in the time domain is equivalent to harmonics in the frequency
domain. We can make an excitation pulse train using a sum of
sinsusoids:
for(m=1; m<=L; m++)
ex[n] = cos(m*Wo*n)
- Note: the Octave script ../octave/phase.m is an example of this if you would
- like to try making a pulse train.
+ Note: the Octave script ../octave/phase.m is an example of this if
+ you would like to try making a pulse train.
The phase of each excitation harmonic is:
first harmonic would advance (pi/20)*80 = 4*pi or two complete
cycles.
- We track the excitation phase of the fundamental (first harmonic):
+ We generate the excitation phase of the fundamental (first
+ harmonic):
arg[E[1]] = Wo*N;
This E[m] then gets passed through the LPC synthesis filter to
determine the final harmonic phase.
+ For a while there were prolems with low pitched males like hts1
+ sounding "clicky". The synthesied time domain waveform also looked
+ clicky. Many methods were tried to improve the sounds quality of
+ low pitched males. Finally adding a small amount of jitter to each
+ harmonic worked.
+
+ The current result sounds very close to the original phases, with
+ only 1 voicing bit per frame. For example hts1a using original
+ amplitudes and this phase model produces speech hard to distinguish
+ from speech synthesise with the orginal phases. The sound quality
+ of this patrtiallyuantised codec (nb original amplitudes) is higher
+ than g729, even though all the phase information has been
+ discarded.
+
NOTES:
- 1/ This synthsis model is effectvely the same as simple LPC-10
+ 1/ This synthesis model is effectvely the same as simple LPC-10
vocoders, and yet sounds much better. Why?
2/ I am pretty sure the Lincoln Lab sinusoidal coding guys (like xMBE
phases are continuous, but not the final phases after filtering
by the LPC spectra). Technically this is a bad thing. However
this may actually be a good thing, disturbing the phase tracks a
- bit. More research needed, e.g. test a synthsis model that adds
+ bit. More research needed, e.g. test a synthesis model that adds
a small delta-W to make phase tracks line up for voiced
harmonics.
+ 4/ Why does this sound so great with 1 V/UV decision? Conventional
+ wisdom says mixed voicing is required for high qaulity speech.
+
\*---------------------------------------------------------------------------*/
void phase_synth_zero_order(
ex_phase[0] += (*prev_Wo+model.Wo)*N/2;
*/
- /* determine jitter offset */
-
- jitter = (1.0 - 2.0*rand()/RAND_MAX);
-
- /* now modify this frames phase using zero phase model */
-
- //ex_phase[0] += (model.Wo)*N - jitter*model.Wo;
ex_phase[0] += (model.Wo)*N;
ex_phase[0] -= TWO_PI*floor(ex_phase[0]/TWO_PI + 0.5);
/* generate excitation */
if (voiced) {
+ /* This method of adding jitter really helped remove the clicky
+ sound in low pitched makes like hts1a. This moves the onset
+ of each harmonic over at +/- 0.25 of a sample.
+ */
jitter = 0.25*(1.0 - 2.0*rand()/RAND_MAX);
Ex[m].real = cos(ex_phase[0]*m - jitter*model.Wo*m);
Ex[m].imag = sin(ex_phase[0]*m - jitter*model.Wo*m);
-
- /* following is an experiment in dispersing pulse energy over
- time, didn't really change sound at all, e.g. mmt1 still
- sounded "clicky. I think this is because this provides
- just a small phase shift between adjacent harmonics.
- However for voiced speech it is the high energy harmonics
- that form pitch pulses, so we need a relatively high phase
- shift between them to disperse pulse energy */
-
- //Ex[m].real = cos(ex_phase[0]*m + model.Wo*m*m*0.3);
- //Ex[m].imag = sin(ex_phase[0]*m + model.Wo*m*m*0.3);
-
- /* following is an experiment to use the phase of a glottal pulse
- (see octave/glottal.m) is an attempt to make mmt1 and hts1 a little
- less "clicky", i.e. disperse the pulse energy away from the point
- of onset. Result was no difference in speech quality, in fact
- no difference at all. Could be an implementation error I guess.
- One again - this model doesnt change phases much between adjacent
- harmonics, so not much dispersion. */
- //b = floor(m*model->Wo*FFT_DEC/TWO_PI + 0.5);
- //Ex[m].real = cos(ex_phase[0]*m + glottal[b]);
- //Ex[m].imag = sin(ex_phase[0]*m + glottal[b]);
-
}
else {
+
/* When a few samples were tested I found that LPC filter
- phase is not needed in the unvoiced case, but no harm in keeping it.
+ phase is not needed in the unvoiced case, but no harm in
+ keeping it.
*/
float phi = TWO_PI*(float)rand()/RAND_MAX;
Ex[m].real = cos(phi);
model.phi[m] = new_phi;
}
- #ifdef CLICKY
- /* Adding a random component to low energy harmonic phase seems to
- improve low pitch speakers. Adding a small random component to
- low energy harmonic amplitudes also helps low pitch speakers after
- LPC modelling (see LPC modelling/amplitude quantisation code).
- */
-
- maxA = 0.0;
- for(i=1; i<=model.L; i++) {
- if (model.A[i] > maxA) {
- maxA = model.A[i];
- }
- }
- for(i=1; i<=model.L; i++) {
- if (model.A[i] < 0.1*maxA) {
- model.phi[i] += 0.2*TWO_PI*(float)rand()/RAND_MAX;
- }
- }
- #endif
}
/*---------------------------------------------------------------------------*\
projects / freetel-svn-tracking.git / commitdiff