[hts-users:03305] Re: How to label phonetic automatically
- Subject: [hts-users:03305] Re: How to label phonetic automatically
- From: Ninh Khanh Duy <ninhkhanhduy@xxxxxxxxx>
- Date: Wed, 16 May 2012 23:45:51 +0900
- Delivered-to: hts-users@xxxxxxxxxxxxxxx
- Dkim-signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20120113; h=mime-version:in-reply-to:references:from:date:message-id:subject:to :content-type:content-transfer-encoding; bh=JHjlC65z8lUR4njJXIX0vGe07HIHLsfUMF6esjSr2Kc=; b=Slm6+AlkfjCHTKWxgrcHwt3IMd9f7cX1AhQ+zbMntCeBhyqKP2s7+BLY7zhdUlEP1/ xD/DYHTGpKgdk2iqyN6D9AKxnE4LAWM9NF7pXsUuu/YTuIMXoetFnIeV9QaQ4NCU0oSn BL6lLjZS8OYbVKXhBMLy3Di32k1qzSeCfLLKLKCTINEJYJI5R5K4dWxMv3aB+hEO7g8t ptcK6o841tubH+iAChveMUj3CKth9g0cCCoZQMM3Wdhp8HEVBtVM7ztn9MCUE55DslK5 C0T5BvYM0VAZFq8ZskNITtxHvOtjWJnWiz6nO7rQTFnCGXxebwQMGrl7DUbi4D62C66V k3Ew==
Hi,
For phonetic labels, you can obtain phone alignments using a speech
recognition engine and well-trained acoustic models for Vietnamese
language.
More importantly, to synthesize acceptable speech quality, you need to
have accurate full-context label for your speech data. Several factors
in full-context label relate to prosodic features of a language and
need to be designed specifically for that language. Considering these
language-specific features, you should develop your own tool or
produce the full-context labels by hand.
Regards,
Ninh
On Wed, May 16, 2012 at 11:13 PM, nguyễn Việt Quốc
<viet.quoc.2569@xxxxxxxxx> wrote:
> hi! thank you!
> that's right that need full-context labels. But i think there are not tool
> to do that.
> easlier, we can label with phonetic, later, we replace phonetic label with
> fullcontext labels.
>
> i see everybody uses some tools to label for recognizing, Can we use it ??
> ^^
> regards.
> Quoc
>
> P/s : Hi Ninh, I am Quoc, Glad to see you, i am Vietnamese, i think you are
> , where are you ,please ?
>
>
> 2012/5/16 Ninh Khanh Duy <ninhkhanhduy@xxxxxxxxx>
>>
>> Hi,
>>
>> To synthesize speech using HTS toolkit, you need not only phonetic
>> labels but also full-context labels associated with speech data. To my
>> knowledge, there is no tool to generate such labels automatically for
>> Vietnamese.
>>
>> Regards,
>>
>> Ninh
>>
>> On Wed, May 16, 2012 at 10:29 PM, nguyễn Việt Quốc
>> <viet.quoc.2569@xxxxxxxxx> wrote:
>> > Dear ,
>> >
>> > I try to synthesis Vietnamese.
>> > it is so lucky, HTS is great tool for this work.
>> >
>> > but, i have some problem for phonetic Labelling utterances to train.
>> > it 's so hard for labeling large Speech Database, because it spents time
>> > so
>> > much and it is not exactly.
>> >
>> > Could you please tell me know whether there is tools for labeling
>> > utterances
>> > automatically ? please.
>> > Can HTS or HTK or Sphinx do it ?
>> >
>> > i need your help so much ,
>> > Thank you.
>> >
>> > --
>> > Nguyen Viet Quoc
>> > Student of Computer Science Department
>> > University of Information Technology, Vietnam National University - Ho
>> > Chi
>> > Minh City
>> > Email: viet.quoc.2569@xxxxxxxxx
>> > Mobile: 01689942711.
>> >
>>
>
>
>
> --
> Nguyen Viet Quoc
> Student of Computer Science Department
> University of Information Technology, Vietnam National University - Ho Chi
> Minh City
> Email: viet.quoc.2569@xxxxxxxxx
> Mobile: 01689942711.
>
- Follow-Ups
-
- [hts-users:03307] Re: How to label phonetic automatically, nguyễn Việt Quốc
- References
-
- [hts-users:03302] How to label phonetic automatically, nguyễn Việt Quốc
- [hts-users:03303] Re: How to label phonetic automatically, Ninh Khanh Duy
- [hts-users:03304] Re: How to label phonetic automatically, nguyễn Việt Quốc