[asterisk-users] Opensource Speech recognition for Asterisk

[Nickolay V. Shmyrev]

Hello Vieri

Indeed, acoustic model is missing. However, if you are really
interested, we can provide you Spanish model for telephone speech in a
week or so. If you can provide some test database with recordings, it
will be even better.

Contact me for details.

--
Nexiwave - Speech Indexing Solution For Call Centers
http://nexiwave.com


2010/8/24, Vieri <rentorbuy at yahoo.com>:
> Hi,
>
> Sorry to drop in on this thread but I'm relatively new to Sphinx and speech
> recognition. I'd like to know if anyone has successfully setup speech
> recognition in Asterisk for Spanish users. Sphinx doesn't seem to have
> Spanish acoustic and language models and I don't think I'll ever have the
> time or know-how to make my own.
> My requirements are similar to the OP's:
> basic "yes", "no", get an 8 digit number, etc.
>
> Actually, "yes" ("si") and "no" work well with the English models. However,
> accuracy is not that great when it comes to recognizing digits zero to nine
> in Spanish.
>
> Thanks for any suggestions,
>
> Vieri
>
> --- On Tue, 8/24/10, Bob Kleiner <bob.kleiner at gmail.com> wrote:
>
>> From: Bob Kleiner <bob.kleiner at gmail.com>
>> Subject: Re: [asterisk-users] Opensource Speech recognition for Asterisk
>> To: asterisk-users at lists.digium.com
>> Date: Tuesday, August 24, 2010, 7:30 AM
>> > Thanks guys. A lot of info here
>> :-)
>> >
>> > I am wondering if anyone followed this and it was
>> working for them:
>> >
>> > http://scribblej.com/svn/
>> >
>> > ???
>>
>> Hello Bruce
>>
>> We successfully deployed it and now saving thousands on
>> commercial ASR
>> ports. It seems users are rather happy with it. The
>> recognition seems
>> pretty accurate. Of course it has it's own limitations but
>> so any
>> other technology. It will not hurt if some of your users
>> will benefit
>> from ASR.
>>
>> > I am not looking for anything fancy. The basic "yes",
>> "no", dialing a
>> > number, asking for agent, etc...out of which probably
>> the hardest is a 10
>> > digit number to be asked to be dialed.
>>
>> Yes, that should work. It also supports JSGF grammars, so
>> you should
>> be able to recognize digit strings easily.
>>
>> And if you want something serious, there are at least two
>> open source products
>> providing ASR over standard MRCP protocol. They also use
>> CMUSphinx, so
>> provide the same accuracy
>>
>> Zanzibar http://www.spokentech.org/writing-speechlets.html
>> Cairo http://www.speechforge.org/
>>
>> Though Cairo is a bit dated.
>>
>> --
>> _____________________________________________________________________
>> -- Bandwidth and Colocation Provided by http://www.api-digital.com --
>> New to Asterisk? Join us for a live introductory webinar
>> every Thurs:
>>
>>    http://www.asterisk.org/hello
>>
>> asterisk-users mailing list
>> To UNSUBSCRIBE or update options visit:
>>    http://lists.digium.com/mailman/listinfo/asterisk-users
>>
>
>
>
>
> --
> _____________________________________________________________________
> -- Bandwidth and Colocation Provided by http://www.api-digital.com --
> New to Asterisk? Join us for a live introductory webinar every Thurs:
>                http://www.asterisk.org/hello
>
> asterisk-users mailing list
> To UNSUBSCRIBE or update options visit:
>    http://lists.digium.com/mailman/listinfo/asterisk-users
>