The knowledge base confirms the following about LiveKit's NC, BVC, and BVCTelephony noise cancellation models: • All three run locally for negligible audio latency and no data is sent to external servers. • NC: Standard enhanced noise cancellation. • BVC: Includes NC, plus removes non-primary voices to improve transcription and turn detection (recommended for 1:1 voice AI). • BVCTelephony: BVC optimized for telephony use cases. All models have negligible impact on audio latency or quality. BVC is suited for voice AI use, and BVCTelephony for telephony. However, the knowledge base does not provide specific quantitative metrics for latency, memory usage, or accuracy differences between the models, nor does it include benchmarks or CPU/GPU requirement comparisons. Waveform comparisons demonstrate improved output compared to standard WebRTC noise suppression, but for detailed technical comparisons or benchmarks, you would likely need to conduct your own performance tests or obtain detailed specs directly from engineering resources. If you need further technical details not listed in the knowledge base, I can escalate your request to a specialist for additional information. • Source: • Source: • Source: Sources: AI noise cancellation | LiveKit Docs | Voice AI quickstart | LiveKit Docs | SIP overview | LiveKit Docs

I don't have the answer you're looking for. You could also try asking your question: • in one of the other Slack channels or • to https://deepwiki.com/livekit/livekit_composite which is trained on all LiveKit source code If you find the answer, please post it here to help others!