[hts-users:03456] Re: new trainig
- Subject: [hts-users:03456] Re: new trainig
- From: Keiichiro Oura <uratec@xxxxxxxxxxxxxxx>
- Date: Sun, 11 Nov 2012 12:50:47 +0900
- Cc: uratec <uratec@xxxxxxxxxxxx>
- Delivered-to: hts-users@xxxxxxxxxxxxxxx
- Dkim-signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20120113; h=mime-version:sender:in-reply-to:references:date :x-google-sender-auth:message-id:subject:from:to:cc:content-type; bh=/DY48ny5yZBCJChMyqR6qt56HvUJU1casqKcKZFuY5s=; b=qsEtUB2czkq+oFH99O++jT0VtwVTXcncaMi3yDBUc3X+dHNTdBL3dY8LS/bI5GtGKL h93iYT77xL2UKySPonpdcRXOQ8V4NPu5lD19kS24jTYhpB002oM081Ft2IqJvJJqYvwT Q2J4UUVGl2XxToDP8ETkz2q2Vf8e3Ii0it6daiS9vp+3ncpRCUCSOfZRdoCB75ItopOG pioX89vrmIfhLQBgFppZC3MXbYy4ypedkBVJBzYdKjZGYcc6zIgbpX4UF8zqFLx1/1hN RnBwR/F0NplkZF3c7W1LLRqmDIzQHQN+w8rc2ER6jFAsg3SVqKsU+l+7zStIZcIRNt/R KcEw==
Hi,
You can train the speaker dependent model with 200 utterances.
Incidentally, the effects of the data size were reported in the following paper.
Jinfu Ni, Hisashi Kawai,
``On the effects of transcript errors across dataset sizes on
HMM-based voices,''
ASJ 2011 autumn, 3-8-6, pp. 339-342, 2011.
Regards,
Keiichiro Oura
2012/11/10 Mohamed Khalil <krichi_moha@xxxxxxxx>:
> hello every body,
> I try to train HTS with new language, my database is small ( 200 files )
> when I train the error is information not enough, I can added other files
> with other person
> what number files( labels) to succeed training HTS?
>
> thanks a lot
>
- Follow-Ups
-
- [hts-users:03457] Re: new trainig, Anar Xocayev
- [hts-users:03458] Re: new trainig, Yongxin Wang
- References
-
- [hts-users:03455] new trainig, Mohamed Khalil