f



Speech rate modeling in synthetic speech

Hi,

I am looking for synthesizers that adequately implement different speech
rates. Synthetic speech at low rate e.g. with MBROLA often sounds like slow
motion speech and at high rate it sounds time compressed and
hyperarticulated. If there are synthesizers that do it correctly HOW do they
do that? Any hints and refs are welcome.

Best regards

Sascha



0
Sascha
6/13/2005 10:50:20 AM
comp.speech.research 1026 articles. 0 followers. Post Follow

2 Replies
808 Views

Similar Articles

[PageSpeed] 47

If someone is answering this question they probably might also answer
this:

http://groups-beta.google.com/group/comp.speech.users/browse_frm/thread/1fbb896947d7819f/6191254eb541d861#6191254eb541d861

Thanks,

~Sarah

0
Sarah
6/18/2005 12:56:03 AM

Sascha Fagel wrote:
> Hi,
>
> I am looking for synthesizers that adequately implement different speech
> rates. Synthetic speech at low rate e.g. with MBROLA often sounds like slow
> motion speech and at high rate it sounds time compressed and
> hyperarticulated. If there are synthesizers that do it correctly HOW do they
> do that? Any hints and refs are welcome.
>
> Best regards
>
> Sascha

In my opinion, the HTS voices in the Festival speech synthesis
system handle speech rates nicely.

To oversimplify:
Higher standard deviation in segment's duration means
that this segment can be stretched or squeezed more.

regards,
  Nicholas

0
Nicholas
6/19/2005 9:36:18 PM
Reply: