[Riqiang]

You can use an arbitrary source for example a pulse which is similar to glottal pulses. If I remember it right, the earliest attempt for speech synthesis is done by vocoder which uses LP. So I think iii is also correct.

But as you can see the source we use is not going to be exactly human voice so the result does not sound good. Vocoder has become popular in music though, for making robot-like sounds.

[Riqiang]

The figure 9.14 in J&M shows a classical cepstrum. Why are we not using this classical cepstrum but the Mel frequency one? We want to isolate the source signal, and it seems to me that the classical cepstrum shown in 9.14 can do just that (harmonics become a single spike corresponding to the fundamental frequency). Although Mel-filterbank also smoothes out the harmonics, it also makes the relationship between harmonics less straightforward (at equal intervals on the Hertz scale)?

[Riqiang]

When talking about the limits of Linear Prediction, Ellis talked about poles and zeros. What do they mean in relation to speech?

[Riqiang]

Hello everyone! Probably we don’t need to know about the details of Fourier transform, but I’m still curious how the scale constant can be calculated by “taking the inner product between the waveform and the harmonic”. How is it done? It seems that the inner product of two waveforms change according to their phase, so how is the scale constant calculated before the phase?

[Author]

Posts