Best Speech Engines on QNX?

QNX Newsgroups (Archives) comp.os.qnx
1

Hello, I am looking for opinions on the best speech recognition engines on
QNX? Are there any good, or even usable ones?

I am looking for fairly small footprint engines that work on
automobile/telematics platforms. Does anyone know if any there speech
engines that have successfully solved the background noise problem in
automobiles? Ideally, we would like an engine that supports a robust
grammer (with garbage wildcards, for example) and perhaps even dictation.

Also, if you know of any good QNX-based TTS engines, please let me know.

Thanks.

(bobk@REMOVETHISvoicebox.com)

2

You will want to look into Cepstral for one. They have a really good
range of voice products and support QNX.

http://www.cepstral.com/

chris


Bob K <bobk@voicebox.com> wrote:

Hello, I am looking for opinions on the best speech recognition engines on
QNX? Are there any good, or even usable ones?

I am looking for fairly small footprint engines that work on
automobile/telematics platforms. Does anyone know if any there speech
engines that have successfully solved the background noise problem in
automobiles? Ideally, we would like an engine that supports a robust
grammer (with garbage wildcards, for example) and perhaps even dictation.

Also, if you know of any good QNX-based TTS engines, please let me know.

Thanks.

(> bobk@REMOVETHISvoicebox.com> )
\

–
Chris McKillop <cdm@qnx.com> “The faster I go, the behinder I get.”
Software Engineer, QSSL – Lewis Carroll –
http://qnx.wox.org/

3

“Bob K” <bobk@voicebox.com> wrote in message
news:b6t1u4$t20$1@inn.qnx.com…

Hello, I am looking for opinions on the best speech recognition engines on
QNX? Are there any good, or even usable ones?

Yes; there are several.

I am looking for fairly small footprint engines that work on
automobile/telematics platforms. Does anyone know if any there speech
engines that have successfully solved the background noise problem in
automobiles? Ideally, we would like an engine that supports a robust
grammer (with garbage wildcards, for example) and perhaps even dictation.

Footprint will be dependent on a number of factors (incl. vocabulary), and
usability for
noisy environments will be highly dependent on calibration of the audio
sub-system and noise suppression software – usually separate from
the VR engine.

We have several vendors who have provided VR technology
for hands-free modules in OEM vehicle telematics programs, including
IBM ViaVoice and Wavemakers (acoustic echo cancellation - AEC), among others
(Fonix, Temic – future availability uncertain since it was acquired by a
tier one supplier;
Clarity for AEC).

AEC dramatically improves the success rate of automated speech
recognition engines in challenging acoustic environments like an automobile.

I don’t know about your particular usage, but ViaVoice for
example is available for specific reference designs due to
the issues cited above (Intel pxa250tmdp, etc.).

Also, if you know of any good QNX-based TTS engines, please let me know.

IBM ViaVoice and Fonix both provide TTS engines as well, and there are
quite a few TTS engines if you include the non-automotive suppliers:

Babel
Cepstral


Thanks.

(> bobk@REMOVETHISvoicebox.com> )