openai_transcription_settings.json

{
    "type": "transcription_session.update",
    "session": {
        "input_audio_format": "pcm16",
        "input_audio_transcription": {
            "model": "gpt-4o-transcribe",
            "prompt": "",
            "language": "en"
        },
        "turn_detection": {
            "type": "server_vad",
            "threshold": 0.4,
            "prefix_padding_ms": 200,
            "silence_duration_ms": 300
        },
        "input_audio_noise_reduction": {
            "type": "near_field"
        }
    }
}