speech recognition vs voice recognition

Started by lucas.bibby, Feb 13, 2017, 08:30 AM

Previous topic - Next topic

lucas.bibby

Braina's speech recognition is pretty impressive and is proving pretty useful.  But speech recognition is one thing, voice recognition is another.

are there any plans to have multiple user interface in Braina?  with voice recognition?  So that say, both my daughter and I could both use Braina, and Braina will recognize each of us by our unique voices...

I like how Braina knows my name and certain other information about me, it advances the illusion of intelligence.  "knowing" multiple users that way would further that illusion considerably.

saurav

Sounds interesting. Forwarded to R&D team. Thank you.

lucas.bibby

#2
Just wanted to mention, I have been experimenting with the Microsoft Speaker Recognition API, I get roughly 333 free calls a day.    And I can achieve it (what i mentioned in OP), sort of.  

I have a couple different VM's running Braina each with different user information profiles, and I use the Microsoft Speaker Recognition API to identify a sound clip when I say "change users... this is Joseph" or "change users... this is Lucas" and then pass microphone control to the VM associated with the results from the API.  

So, "change users" calls a Braina custom command (because Braina has control of the mic) which initiates the script to take input for the API which takes control of the mic and once that is listening, I give it a sound sample "this is Joseph" or "this is Lucas" and it then passes that sound clip to the API, gets a result, returns the result and a confidence level, then my script opens the appropriate VM and tells that VM's Braina to "start listening" which makes that Braina take control of the mic.

So far, I have only been able to test it with my own voice, and me speaking in an accented voice (like voice acting, kind of).  and it is really really slow due to the time to execute Braina command, run that script, get the sound sample, pass the sound sample and retrieve results then open the appropriate VM, so I cannot use it inline, as in in a sentence. Or change users based on voice alone without initiating "change users" command.... But it works.  Next time I have my daughter, I will test it with her voice.

lucas.bibby

I was able to speed it up a bit by eliminating the VM's and having the script Update Profile in Braina.