Repository Issue Activity (beta)

#1987 Unable to detect Overlaped speakers

#1974 Is torchaudio 2.9 supported?

#1570 Can I download a separate audio file for each speaker?

#2050 SpeakerDiarization computes each chunk's embeddings once per speaker slot, although the WeSpeaker mask only enters at the pooling stage — ~2x end-to-end speedup available

#1979 it has some problems on README

#2044 Loading speaker-diarization-3.1 requires the gated speaker-diarization-community-1 repo for a PLDA the pipeline never uses

#2034 PYANNOTE_CACHE silently overrides HF_HUB_CACHE, causing file-not-found errors in offline/containerized deployments

#2042 Discrete DER reports 0.0 on long recordings because frame counts saturate float16

#180 How to prepare dataset for pyannote-audio.

#1969 Does not work on 5-second file

#2000 Remote Code Execution via unsafe class instantiation from config.yaml in Pipeline.from_pretrained

#1966 Manual diarization pipeline (segmentation + embedding + clustering) unstable on long audio (>1 min) but works on short audio

#1964 pyannote.audio 4.0.3 pyannote/speaker-diarization-precision-2 An error occurred

#1942 Question about using pyannote/speaker-diarization-3.1 in C++

#2032 Support for batch inference breaks system

#2021 CI/CD Actions on the release, PyPI publish and docs deploy paths are not pinned to commit SHAs

#1952 AttributeError: module 'torchaudio' has no attribute 'AudioMetaData'

#2019 Remote Code Execution via Attacker-Controlled config.yaml Class Loading

#1961 Error forming embeddings for small duration audio: Found speech but no valid (non-NaN) speaker embeddings could be generated.

#1941 numpy.NAN deprecated with numpy 1.26 and pyannote 3

#1861 UserWarning: std(): degrees of freedom is <= 0

#1940 Could not load libtorchcodec

#1955 Performance issue on the embeddings step

#1965 pyannote.audio 4.0.3 pyannote/speaker-diarization-precision-2 An error occurred

#2006 pyannote-audio==3.3.2` is uninstallable due to `lightning` package quarantine on PyPI