How per-participant audio capture works
In a video conference, every participant's microphone already produces its own stream; the platform mixes them together for playback. Per-participant capture taps those streams before or instead of the mix. A meeting bot inside the call subscribes to each participant's audio individually and exposes them as separate tracks, either as live streams or as separate files after the meeting. The result is a set of clean, participant-labeled tracks that carry both the audio and the identity of the person who produced it.
Why separate tracks beat a mixed recording
A single mixed track discards information that is expensive or impossible to recover later:
- Speaker attribution becomes exact, since each track maps to a known participant
- Overlapping speech stays intelligible because voices never collide on one channel
- Transcription improves, as engines process one clean voice at a time
- Volume and quality can be normalized per person
- Analysis such as sentiment or talk time runs per speaker without diarization errors
Use cases that depend on per-participant audio
Conversation intelligence platforms rely on isolated tracks to measure talk ratios and interruptions accurately. Interview and user research tools need clean attribution of questions and answers. Podcast-style production workflows remix individual tracks. Voice AI teams use isolated speech as high-quality training or evaluation data, and real-time agents use it to know exactly who is addressing them.
Practical considerations
Not every capture method offers it: native platform recordings usually produce only a mixed track, so per-participant audio typically requires bot-based capture through an API. Expect one stream or file per attendee, so storage and processing scale with participant count, and plan for edge cases such as several people sharing one conference-room microphone, where diarization within a track is still useful.
Per-participant audio in practice
MeetStream bots deliver separate audio streams per participant in roughly 200ms over WebSocket, alongside MP4 recordings and speaker-identified transcripts. One API call with the meeting link starts capture on Zoom, Google Meet, or Microsoft Teams. The MeetStream docs describe the stream format. See the meeting bot API page to go deeper.
Frequently asked questions
Does per-participant audio remove the need for diarization?
Largely, yes. Each track already belongs to a known participant, so attribution is direct and does not rely on voice clustering. Diarization still helps within a single track when multiple people share one device, such as a conference room.
Can I get per-participant audio from a normal cloud recording?
Usually not. Built-in platform recordings deliver one mixed track. Separate tracks generally require capturing inside the meeting with a bot that subscribes to each participant's stream individually.
Is per-participant audio available live or only after the call?
Both, depending on the provider. Live separate streams enable real-time transcription and agents, while separate post-call files suit editing and analysis workflows. Some APIs offer the two together from the same bot.
Put the vocabulary to work
One API to join, record, stream, and transcribe meetings on Zoom, Google Meet, and Microsoft Teams.