FreeSwitch Deepgram Voice AI Agent Integration

Job ID: 39841633

Budget: $30 – $250 USD

We are building a real-time voice agent using Deepgram's WebSocket API and need an expert in Freeswitch to help us bridge our current call flow to the Deepgram voice agent.
 
Requirements:
- Proven experience with Freeswitch core and module development
- Experience with `mod_audio_fork` and `mod_audio_stream`
- Deep understanding of SIP/RTP/media flows
 
What you will do:
- Connect our existing Freeswitch server with Deepgram’s WebSocket-based voice agent using `mod_audio_fork` and `mod_audio_stream`, we need both to be configured. 


Enable seamless, real-time, bi-directional audio between the caller and the voice agent
.

Stream audio to Deepgram in real-time and handle incoming transcription/command messages.

Maintain high availability and low latency across multiple concurrent sessions.

Ensure voice agent can:
  - Execute in-call commands like:
    - End the call
    - Transfer the call to a human agent
    - Trigger DTMF or SIP-based routing actions
    - Play custom messages or handle call hold
-

Implement and maintain synchronized call recording (merged caller and agent audio in a single file) without disrupting existing call flows
- Set up retry/reconnect mechanisms on WebSocket failure
- Collaborate with our backend team to ensure correct message routing between Freeswitch and voice agent via secure WebSocket communication
- Optimize system performance and provide documentation for deployment and maintenance
 
Our FreeSWITCH is the latest version.

This project is by bid but we are also looking for a new telecom engineer who does good work at a good price, we need a lot of other things done and our OpenSIPs updated month many other things, if we work well together there will be plenty more work, we are a small startup that is getting ready to launch soon.

Thank you!
Related categories: VoIP FreeSwitch Audio Processing SIP