[Subject Prev][Subject Next][Thread Prev][Thread Next][Date Index][Thread Index]

[hts-users:04335] Re: bad voice output for test sentences



On 22 Nov 2015, at 21:39, Erica Cooper <ecooper@xxxxxxxxxxxxxxx> wrote:

Thanks very much for the advice.  It is true that the original data was 16kHz and then up-sampled to use with the demo.

I’m sure you know this, but sometimes people forget the basics: upsampling will result in a signal with no energy between the original Nyquist frequency (8 kHz in this case) and the new Nyquist frequency (24kHz in this case). That means a pretty tricky spectral envelope for the cepstral estimation to fit to.

Simon

The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.

References
[hts-users:04325] bad voice output for test sentences, Erica Cooper
[hts-users:04326] Re: bad voice output for test sentences, Keiichiro Oura
[hts-users:04327] Re: bad voice output for test sentences, Erica Cooper
[hts-users:04329] Re: bad voice output for test sentences, Keiichiro Oura
[hts-users:04334] Re: bad voice output for test sentences, Erica Cooper