AssistantsCustom transcriber

Gradium

Use Gradium speech-to-text as a custom transcriber in Vapi

Gradium builds real-time audio models, including streaming speech-to-text with semantic voice activity detection. You can connect it through a custom transcriber endpoint: a small WebSocket bridge that forwards call audio to Gradium and returns transcripts to Vapi.

A Gradium account and API key are required.
1

Build the bridge

Gradium’s Vapi integration guide has the complete bridge, including end-of-turn detection and reconnecting on long calls.

Two things matter most: forward only channel 0 of the interleaved audio, or the agent transcribes its own speech; and send partial transcripts as well as final ones, which Vapi needs for barge-in.

2

Point your assistant at it

Host the bridge somewhere Vapi can reach over wss, then reference it with a custom credential so Vapi can authenticate to it:

1{
2  "transcriber": {
3    "provider": "custom-transcriber",
4    "server": {
5      "url": "wss://your-server.com/vapi/transcriber",
6      "credentialId": "cred_transcriber_auth_123"
7    }
8  }
9}

Start a call and your assistant transcribes with Gradium.

Troubleshooting

SymptomFix
The agent transcribes its own speechBoth channels are being forwarded. Send only channel 0 from the interleaved PCM
Turns get swallowed, barge-in stops workingSend partial transcripts as well as final
Final transcripts lose the last word or twoFinalize on Gradium’s flush acknowledgement, not on the VAD threshold crossing. See Gradium’s Vapi guide
The assistant replies before the caller finishesVapi’s endpointing is racing Gradium’s VAD. Tune start and stop speaking plans, or hand the decision to Gradium with a custom endpointing model
Gradium Vapi guide

The full bridge, tuning, and configuration reference.

Gradium text-to-speech

Use Gradium voices as a custom voice.