feat: speaker diarization with voice-print identification - #3

Merged

Pinous merged 5 commits into

feat/models-command-and-cifrom

feat/speaker-diarization

Jul 31, 2026

Commits on Jul 31, 2026

feat: speaker diarization with voice-print identification

Pinous
and
claude
committed
1515f2e
fix: identify known speakers without asking for a flag

Pinous
and
claude
committed
be5a702
feat: detect the speaker count automatically

Pinous
and
claude
committed
fdfd889
fix: bound the speaker-count probe to the longest turns

Pinous
and
claude
committed
63bcf67
fix: sample one turn per label so quiet speakers reach the probe

Pinous
and
claude
committed
a9c6bfb