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

[hts-users:00666] Re: question about training new voice



hi Zen,
i was appreciated for your reply,  and i think i will try to build chinese HTS voice, 
after i serach hts-user ML, i quess *.utt were generated using festvox(clunits voice building process), 
right ? i want to make sure. So i can generate *.utt using my own voice database.
 
2007/4/24, Heiga ZEN (Byung Ha CHUN) <zen@xxxxxxxxxxxxxxx>:
Hi,

Ching-Hsien Lee wrote (2007/04/24 1:35):

> According to "Training a new voice"
> article(http://hts.sp.nitech.ac.jp/hts-users/spool/2006/msg00213.html )
> in that :
> ----------------------------------------------------------------------------------------------------------------------------------------------
> ...
> (2)    Downloaded two male voices
>  a.
> <http://www.speech.cs.cmu.edu/cmu_arctic/packed/cmu_us_bdl_arctic-0.95-release.tar.bz2 >
> http://www.speech.cs.cmu.edu/cmu_arctic/packed/cmu_us_bdl_arctic-0.95-release.tar.bz2
>  a.
> http://www.speech.cs.cmu.edu/cmu_arctic/packed/cmu_us_rms_arctic-0.95-release.zip
> (3)     Converted audio files from WAV format to RAW.
> (4)     Copied utt files to the training directory
>  ../festival/utts/*.utt à <training>/data/utts/
> ...
> ----------------------------------------------------------------------------------------------------------------------------------------------
>
> the first question is that accroding above list (4)  "Copied utt files
> to the training directory"
> how to generate *.utt ? does use festvox, building clunits voice units
> database to generate my *.utt ?

The above article discussed about how to train additional CMU ARCTIC voices.
I guess you are trying to construct Chinese voices.
Discussions about "creating new voices for new languages" has already been discussed several times.
Please search again.

I also recommend you to read

K. Tokuda, H. Zen, A.W. Black, An HMM-based speech synthesis system applied to English, Proc. of 2002 IEEE SSW, Sept. 2002.

This paper describes what we did to apply HTS to English.
You can find this document on HTS Publications page.


> and besides the *.utt, what do the content of
> /HTS-demo_CMU-ARCTIC-SLT/data/

See README on this directory.

>  i need to generate myself ?

Most of them (except raw, utts, questions, win) are automatically generated from Makefile.

Regards,

Heiga ZEN (Byung Ha CHUN)

--
------------------------------------------------
Heiga ZEN     (in Japanese pronunciation)
Byung Ha CHUN (in Korean pronunciation)

Department of Computer Science and Engineering
Nagoya Institute of Technology
Gokiso-cho, Showa-ku, Nagoya 466-8555 Japan

http://www.sp.nitech.ac.jp/~zen
------------------------------------------------



Follow-Ups
[hts-users:00667] Re: question about training new voice, Heiga ZEN (Byung Ha CHUN)
References
[hts-users:00663] question about training new voice, Ching-Hsien Lee
[hts-users:00665] Re: question about training new voice, Heiga ZEN (Byung Ha CHUN)