AMI Meeting Corpus

~100 h of recorded English meetings with synchronized audio, video, and rich annotations including transcripts and speaker labels.

• Task: asr, speaker
• Languages: en
• Hours: 100
• Domain: meetings
• License: CC BY 4.0
• Homepage: https://groups.inf.ed.ac.uk/ami/corpus/
• Paper: https://dl.acm.org/doi/10.1007/11677482_3

Recommendation

A standard benchmark for meeting-domain ASR, speaker diarization, and overlapping speech, with headset and far-field mic conditions. Choose it when you need multi-party conversational audio. Only ~100 h and largely non-native English, so it is small for training from scratch.

Getting the data

Obtain from the dataset homepage.

Also accessible via OpenSLR (openslr.org/16) and the HF mirror edinburghcstr/ami.

Suggested processing

A recommended VoxKitchen pipeline ships in the repository at voxkitchen/templates/pipelines/speaker-analysis.yaml — run it with vkit docker run.