Multichannel Meetings: 10 Hours of Co-Located Multi-Speaker Recordings

GroundTruth Multi-Channel Meeting Audio Dataset contains approximately 10 hours of co-located, multi-speaker meeting recordings. Each meeting was captured simultaneously via a room microphone and individual close-talk lapel microphones worn by each participant. The dataset was created by 'ground-truth' and was last updated on Hugging Face in April 2026.

Use Cases

• Training speech separation models based on the availability of individual close-talk recordings.
• Developing speaker diarization algorithms based on multi-speaker meeting audio.
• Benchmarking audio enhancement techniques using the paired room and lapel microphone recordings.
• Studying conversational dynamics in co-located meetings based on the multi-channel audio.

Strengths

• Contains approximately 10 hours of audio data.
• Provides simultaneous recordings from both room and individual lapel microphones.
• Includes structured metadata describing speakers.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• The dataset description is incomplete, referring users to an external page for full details.

Provenance

Source

ground-truth

Collection Method

Recordings of co-located meetings captured via built-in room microphone and individual lapel microphones.

Time Range

null

Freshness

Last updated 2026-04-27 07:55:58; freshness should be verified.

Geography

null