Dear Doctor Heiga,
Do you have HMMs for your target language?
i didn't yet generate the corresponding HMMs model for each phoneme,i mean the segmentation from the speech file ...!
If yes, run HVite with -a option using these HMMs.
you mean that the HVite -a option extract the sequence of HMMs model phonemes into sequence of text phonemes (Recognition?)
If no, you can use flat-start training (start from HERest phase).
It does not require phone boundaries.
in the HTS i found that u use the HCompv and HERest(flat starting) while u use in the data prepation lableing files the corresponding boundaries for each phonemes (i have a conflict)
For both cases, you need the corresponding phoneme sequence for each utterance.
You mean by the corresponding phoneme sequence the recorded sentences
sorry for all those questions if it was stupid coz i'am a beginner to understand the Mecahnism of the HTS engine
Please refer HTKBook section 3.2 for detail.
i'am already checked this belonged section
Many Thanks To you
Tamer Fares