Integrating speech speed up code into emacspeak and/or Android stack

[Bill Cox <waywardgeek@xxxxxxxxxxx>]

I've been working for several weeks on various techniques for speeding
up voice by very high factors, well over 2X, and have finally
succeeded.  I've created a very simple GLP library that implements my
new algorithm, which I'm calling Sonic.  It is optimised for high rate
increases, unlike previous algorithms.  In my opinion, voxin sounds as
good, and maybe better, when sped up using with Sonic, instead of
directly generating high speed speech with voxin.  Espeak, in my
opinion, sounds much better at 2.6X speed up when sonic is used as
opposed to using 'espeak -s 450'.  Also, sonic enables espeak to speak
at rates much faster than espeak's current limit of about 2.6X.

I believe the most beneficial places to include sonic support are the
speech back-ends in the various open-source operating systems.  This
includes the emacspeak speech server, and the Android back end voice
stack.  I currently listen to the Loquendo voice on my Nexus One, and
I really like the voice.  However, when played at the painfully slow
rate Android calls "very fast", Loquendo sounds highly distorted.
With Sonic integration, we can enable Loquendo to play at 2X, 4X, or
faster, without nearly the distortion evident today.

Sonic is super simple, the core of the algorithm is just around 170
lines of code, and can be found at the bottom of sonic.c.  There are
some sound samples in the samples directory which show the quality of
sped up speech.  A tar-ball of the project can be downloaded here:

    http://vinux-project.org/sonic-0.1.3.tar.bz2

It's currently hosted in git at vinux-project.org, and released in the
Vinux/Lucid PPA on launchpad.net.

Bill

-----------------------------------------------------------------------------
To unsubscribe from the emacspeak list or change your address on the
emacspeak list send mail to "emacspeak-request@xxxxxxxxxxx" with a
subject of "unsubscribe" or "help".