[asterisk-users] Asterisk- speech to text(Voicemail totext message)

[Danny Nicholas]

-----Original Message-----
From: asterisk-users-bounces at lists.digium.com
[mailto:asterisk-users-bounces at lists.digium.com] On Behalf Of Justin
Sherrill
Sent: Thursday, September 23, 2010 12:07 PM
To: Asterisk Users Mailing List - Non-Commercial Discussion
Subject: Re: [asterisk-users] Asterisk- speech to text(Voicemail totext
message)



>From: asterisk-users-bounces at lists.digium.com
[mailto:asterisk-users-bounces at lists.digium.com] On 
>Behalf Of Danny Nicholas
>Sent: Wednesday, September 22, 2010 5:04 PM
>To: 'Asterisk Users Mailing List - Non-Commercial Discussion'
>Subject: Re: [asterisk-users] Asterisk- speech to text(Voicemail to text
message)
>
>FWIW, the current state of Speech-to-text will let you do a 70-95% accurate
translation of 
>incoming voicemails depending on clarity/dialect/training.  Also depends on
language of 
>"native" speakers.  For 100% reliability, this still requires Human
intervention.

I'd like to do this too.  Poking around, it looks like res_speech.so is the
library to enable it, but an actual separate program to convert from voice
to text is needed, like Sphinx or VXI?  I haven't found anything yet that
describes how to connect it to voicemail.  Examples are welcome, if anyone
has one to point at/paste.

Looking at Sphinx and the available documentation, I think these things to
be true.
#1 - res_speech.so isn't necessary since Sphinx operates as a external
module as opposed to the resident modules of Vestec and Lumenvox.
#2 - Didn't really find a good "on-the-fly" example of processing the file
as it came in.