Specch capture

[ankur.taunk]

I want to capture Users Speech i.e. I need to pragmatically capture TCP stream of user speech in IVR (may it be in Java or Java Script etc.)

Any thought how can I do that ? In which component in Cisco's IVR architecture ?

[david.macias]

I would do it on the Nuance side. You can set Nuance to capture every utterance.

david

[ankur.taunk]

Meaning integrating Nuance component with Cisco's IVR component. Or go with nuance IVR altogether ?

[david.macias]

How are you doing speech now?

david

[ankur.taunk]

We use Record element provided my CVP. Store the audio at a location and convert that audio to text. Not a optimized approach - lot of latency, IO, CUP cycles etc. Thus, trying to find a approach where direct TCP stream can be fed to any Speech API's; to make it close to real time.

Thoughts ?

[david.macias]

Not what I was imagining. I can't think of any way to do at the packet level and even if you could you would still need to "hear it" to turn it to text. How are you doing the audio to text piece now?

david