[hts-users:02378] Synchronisation of articulatory and acoustic flows in HTK ?
In the context of acoustic-articulatory HMMs consisting of synchronous
acoustic vectors (MFCCs) and articulatory vectors (XY EMA coordinates),
we would like to know the convention for positioning the time windows
(often 24 ms) used to generate the MFCC coefficients.
Specifically, can we consider that for a given EMA vector, the
associated MFCCs have been computed from a windows that is centered on
this instant?
Regards,
Atef Ben Youssef
- References
-
- [hts-users:02375] Building HMM voice, Dipesh Dasgupta
- [hts-users:02377] Re: Building HMM voice, Keiichiro Oura