Talk Interface

Real-time voice conversation with your AI models via WebRTC

Disconnected

Note: Select a pipeline model below and click 'Connect' to start a real-time voice conversation. The pipeline model includes VAD, transcription, LLM, and TTS components. Your microphone audio streams continuously; the server detects speech and responds automatically.

Pipeline Model Select a pipeline model {{ range .PipelineModels }} {{.Name}} {{ end }}

VAD

Transcription

LLM

TTS

#Session Settings

InstructionsYou are a helpful voice assistant. Your responses will be spoken aloud using text-to-speech, so keep them concise and conversational. Do not use markdown formatting, bullet points, numbered lists, code blocks, or special characters. Speak naturally as you would in a phone conversation. Avoid parenthetical asides, URLs, and anything that cannot be clearly vocalized.

Voice

Transcription Language

Conversation will appear here...

ConnectTest ToneDiag Disconnect

Audio Diagnostics

Waveform (time domain)

Spectrum (FFT)

Peak Freq

THD

RMS Level

Sample Rate

Packets Recv

Packets Lost

Jitter

Concealed