LiveKit Community

if i am using realtime model from google, what is the sampling rate, chunk size and output frame bytes

Thanks for reaching out, Tarush. I'm an AI assistant and I'm here to help. Give me a moment while I research this topic.


For the Google real-time model in LiveKit, here are the audio specifications:

• *Sampling Rates:*
    ◦ Input audio sampling rate: 16,000 Hz
    ◦ Output audio sampling rate: 24,000 Hz
• *Chunk Size:* Input chunk size is 50ms, which equals 800 samples per channel (calculated as `INPUT_AUDIO_SAMPLE_RATE // 20` where the sample rate is 16,000 Hz).
• *Audio Channels:* Both input and output use 1 channel (mono).
However, the exact output frame bytes for the Google real-time model are not explicitly documented. If you need further clarification or specific details about the output frame bytes, I can escalate this to a specialist for more precise information.
 Sources: <https://github.com/livekit/agents/blob/9c76e2e67f9817dc63b5db07540bec3ef1b55ae1/livekit-plugins/livekit-plugins-openai/livekit/plugins/openai/realtime/realtime_model.py#L1-L67|livekit-plugins/livekit-plugins-openai/livekit/plugins/openai/realtime/real...> | <https://github.com/livekit/agents-js/blob/20f5ef49fe9051419af3c3f502253d4a093bbbbf/plugins/openai/src/realtime/realtime_model.ts#L1-L84|plugins/openai/src/realtime/realtime_model.ts>