HMM for Synthesis

[Sambit P]

Considering how HMMs are generative models, can we not use HMMs trained using some data to generate MFCCs and then convert that into speech? Can that be used for synthesis?

[Simon]

Yes, HMMs can be used for speech synthesis. We would not use MFCCs because they are not invertible, due to the filterbank. We might use something closer to the true cepstrum.

[Author]

Posts