get audio stream from voip to pass to STT

[shihabkb@gmail.com]

We are building an app that get the audio stream (live stream and recorded) from voip call and pass it to a speech to text engine. The speech to text engine accept websocket as input. Is there any way that we can get the realtime audio/recorded audio from cisco voip calls?

[Jonathan Schulenberg]

Sure. Use the Call Recording framework in CUCM to have it fork a copy of the audio to you. The phone will even send separate streams for the calling/called participants.

If you need to selectively invoke this you would need a TAPI/JTAPI integration to get the CTI event and trigger the recording. The easier approach is configure automatic recording and CUCM will send your app a SIP INVITE for every call immediately when it starts.
https://www.cisco.com/c/en/us/td/docs/voice_ip_comm/cucm/srnd/collab12/collab12/recordng.html

https://www.cisco.com/c/en/us/td/docs/voice_ip_comm/cucm/admin/12_0_1/featureConfig/cucm_b_cucm-feature-configuration-guide_1201/cucm_b_cucm-feature-configuration-guide_1201_chapter_01010.html

PS- ASR typically requires G.711 or better to work. Don’t overlook your Region configuration in CUCM.