David, Again, what exactly are your requirements?
Do you require to identify a word ? Do you require to understand the whole sentence ? Do you need to know that "David" is "David" and not someone else ? The three types of scenarios here are three different types of systems, with three different types of products to use, and the voice signature has it's own companies in that field. To recognize a person, there are few companies in the world that does it, and it costs tons of money, and depends on so many levels. The thing is, that telephony (e.g. land line and sip trunk to land line) always works in 8KHz (usually g711 and g729 in codecs), and that makes things not very good to the subject, however codecs such as OPUS works in Full HD and provides a lot more data to use, but not supported in many places. To understand a sentence, there are voice recognition services, such as Nuance's, that afaik, does not capable in supporting Hebrew at all. Last, you have keyword spotting, for example the way Pago works. You say a name of a city, and it try to figure out pattern that is close enough to that city. It cannot understand all the words, but specific words that is was programmed to find out. For example the Israeli Police uses it for getting your ID number when you want to know about speeding tickets in the IVR, and it doesn't work in all cases. I do not know of any open source solution for any of this type of solutions. AFAIK, Asterisk does not support it by default, but requires plugins depends on the system, or to send recordings to external system. Freeswitch (another open source PBX) does support some of the protocols, but not all, again, depends on what the exact solution is. There is also YATE (another open source PBX), that I have never used, so I cannot say, It's too hard to answer more then that, without you explaining your actual needs. Ido On Mon, Jan 19, 2015 at 5:41 PM, David Ronkin <[email protected]> wrote: > I know some companies sell this but, lets take Asterisk or similar, can > anyone give me a short overview, please: > > > 1. how machure the speech recognition support in a FOSS ? which one is > best to start > 2. i need basic voice/speech recognition ability, to distingish a man in > a short possible time > 3. what about hebrew (other langs) support? > > Thanks! > David > > > 2015-01-19 8:17 GMT+02:00 ik <[email protected]>: > >> Hi, >> >> I'm not that person, but you're message is too cryptic. >> >> There are voice signature systems, there are key words spotting, and >> attempt to decipher meaning of words systems, each system contains >> difference the requirement of what is required. >> >> There are also human language issues to know about, each language has >> it's own issues to deal with. >> >> I do not know of any good open source software for any of that. >> >> Ido >> On Jan 18, 2015 10:54 AM, "David Ronkin" <[email protected]> wrote: >> >>> Hi all >>> >>> I'm looking for a specialist in voice recognition solutions ( based >>> Asterisk or similar, FOSS preferable of course ) for a long term freelance >>> contract. >>> Only people with relevant experience please contact me off the list david >>> at ok2go.co.il >>> >>> David >>> http://www.ok2go.co.il/ >>> >>> >>> >>> >>> >>> >>> -- >>> בברכה, >>> דוד רונקין >>> >>> *נא בקרו בבלוג של*י: http://dronkin.blogspot.com >>> >>> >>> >>> >>> >>> >>> _______________________________________________ >>> Linux-il mailing list >>> [email protected] >>> http://mailman.cs.huji.ac.il/mailman/listinfo/linux-il >>> >>> > > > -- > בברכה, > דוד רונקין > > *נא בקרו בבלוג של*י: http://dronkin.blogspot.com > *הערוץ של מאיר*: http://www.youtube.com/user/ronkinim > > > > >
_______________________________________________ Linux-il mailing list [email protected] http://mailman.cs.huji.ac.il/mailman/listinfo/linux-il
