Automated Interpreter

Ready to interpret

Choose Start to connect your microphone and begin a live translated conversation.

Not connected

Sandbox session

Interpreter settings

Connection

WebSocket endpoint *

The Edge generic streaming endpoint (PCM16 · 16 kHz).

Tenant ID *

Auto-generated; edit if you have a specific tenant. Sent as tenant_id.

Languages

Spoken language

Translated language

Translated speech is spoken back in the target language.

Audio devices

Microphone Speaker

Translated audio plays through this output.

Preview voice (mock only)

Speed 0.95×

The live build speaks with the platform's neural TTS streamed from the service — this preview voice only affects the mock.

Prevent speaker feedback Pause microphone capture while translated audio is playing.

⚙ Advanced — ASR model

Speech-to-text engine

Auto lets the platform pick per language. Overriding is sandbox only — it works when the Edge has GENERIC_ALLOW_PROVIDER_OVERRIDE=true, otherwise it's ignored.

{} Session details

call_id— sample_rate16000 Hz · PCM16 mono outputpcm16 (translated TTS) stateidle