[PLUG] Smoothing audio

Quentin Hartman qhartman at gmail.com
Tue Apr 29 15:36:27 UTC 2008


On Tue, Apr 29, 2008 at 8:19 AM, Roderick A. Anderson <raanders at acm.org>
wrote:

> I haven't done any serious research but was hoping to get some ideas on
> what terms to use or applications to consider before going further.
>
> BACKGROUND:
>
> I use Cepstral's "Text To Speech" (TTS) engine to generate messages for
> use by Asterisk.  Though the engine and selected voice works very well
> there are still a few rough spots in speech generated.  Cepstrals TTS
> engine does support a mark-up language but I'm thinking there should be
> some application that would (automagically?) allow me to smooth the
> glitches out.
>
> I've tried Audacity but didn't notice any option to do this.  Most
> probably I didn't recognize it for what it is.
>
> Any suggestions of what to search for?
>

First I think you need to better define what you mean by "smooth". Are there
pitch problems? Cadence problems? Inflection problems? If Cepstrals supports
mark-up to adjust pronunciation I'd attack it from that direction, you're
likely to have much better results. Manipulating audio is definitely an art.
I doubt you will find anything that  will automagically do what you are
after, unless there are clear and well defined aberrations that are simple
to isolate. It doesn't sound like that's the case though.

-- 
-Regards-

-Quentin Hartman-



More information about the PLUG mailing list