[hts-users:03653] Re: Improving the vocoder performance
- Subject: [hts-users:03653] Re: Improving the vocoder performance
- From: Keiichiro Oura <uratec@xxxxxxxxxxxxxxx>
- Date: Wed, 6 Mar 2013 17:04:39 +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:x-received:sender:in-reply-to:references:date :x-google-sender-auth:message-id:subject:from:to:cc:content-type :content-transfer-encoding; bh=2nDgTEb32w0ZkfklLnm/ZvrUD4w3l+6vxy1UnNqHuHo=; b=n0M4DKY9sGQCsAg8UPnRMkegrSEa6F1qpI1TFGYes1o4wMp6T0DnhqlYJ2eGqj0xG+ R6hlIjqJS0fNNUHg4+CM7VWEIDIAINqewP4ATduYx/3P73p9R9T6rh5ybgI+XJROsySb F6niL/exGqL+xvTu2myTSMkPkMb6qV3+Gw8tq6qhOt27fTnika5GG6wmsBv1V1/9VfNK 6KPjFq6KsxGVc4L2ls3MaG7NvF4oSS3lT96pLcCptJ6+Uy+ld/BdVL9cq7McI6b4cejg F0zls7zxw8YmDGFlhPUleWj6qcOrTnrL2LpC15si6UYoRtMDGOpOgVycWsTeeSACa+96 3WMA==
Hi,
How about following options for speed up?
Sampling rate: 16000Hz -> 8000Hz (down-sampling of training data is required)
Number of states: 5 -> 3
Frame shift: 5ms -> 10ms
Audio bufffer: 800 (synthesized speech samples should be threw into
audio device directly)
GV: off
Postfilter: off
Alpha: 0.0
Feature: 12th order LSP and Log F0
MDL thresshold: 1.0 -> 3.0
Regards,
Keiichiro Oura
2013/3/5 pramod kumar <pramod2001.india@xxxxxxxxx>:
> Sampling rate: 16000
> Frame shift: 5ms
> Audio buffer: 1600 (can be changed)
>
> By turning off the post filtering and un-necessary conditional statement, am
> able to reduce the discontinuity little bit. Still, looking for other
> options to improve the speed.
>
>
>
> On Mon, Mar 4, 2013 at 9:35 PM, Keiichiro Oura <uratec@xxxxxxxxxxxxxxx>
> wrote:
>>
>> Hi,
>>
>> Please, let's me know the sampling rate, frame shift, audio buffer
>> size (used in OpenSLES), etc.
>>
>> Regards,
>> Keiichiro Oura
>>
>>
>> 2013/3/2 pramod kumar <pramod2001.india@xxxxxxxxx>:
>> > As mentioned in the paper, am already using 12th order LSP. Still, there
>> > is
>> > a latency in the play back continuation.
>> >
>> >
>> > On Sat, Mar 2, 2013 at 5:55 PM, Csapo Tamas Gabor <csapot@xxxxxxxxxxx>
>> > wrote:
>> >>
>> >> Hi,
>> >>
>> >> I suggest you to read this article, there are a couple of steps to
>> >> improve
>> >> performance on mobile:
>> >>
>> >> B. Tóth, G. Németh, Optimizing HMM Speech Synthesis for Low-Resource
>> >> Devices, JOURNAL OF ADVANCED COMPUTATIONAL INTELLIGENCE AND INTELLIGENT
>> >> INFORMATICS 16:(2), 2012, pp. 327-334.
>> >>
>> >> http://speechlab.tmit.bme.hu/toth/downloads/tothb-2012-jaciii.pdf
>> >>
>> >> Best,
>> >> Tamas
>> >>
>> >> 2013.03.01. 18:48 keltezéssel, pramod kumar írta:
>> >>
>> >>> Hi,
>> >>>
>> >>> We would like to use HTS (2.3 version) voices using hts_engine in
>> >>> Android. Everything is working fine.
>> >>>
>> >>> Since, mobile we are targeting basic smart phones (which comes with
>> >>> 512
>> >>> MB RAM), there is a delay in the audio output streaming.
>> >>>
>> >>> What are the possible steps to improve the vocoder performance,
>> >>>
>> >>> --
>> >>> Thanks & Regards,
>> >>> Pramod.
>> >>
>> >>
>> >
>> >
>> >
>> > --
>> > Thanks & Regards,
>> > Pramod.
>>
>
>
>
> --
> Thanks & Regards,
> Pramod.
- Follow-Ups
-
- [hts-users:03661] Re: Improving the vocoder performance, pramod kumar
- References
-
- [hts-users:03643] Improving the vocoder performance, pramod kumar
- [hts-users:03644] Re: Improving the vocoder performance, Csapo Tamas Gabor
- [hts-users:03645] Re: Improving the vocoder performance, pramod kumar
- [hts-users:03647] Re: Improving the vocoder performance, Keiichiro Oura
- [hts-users:03648] Re: Improving the vocoder performance, pramod kumar