modelscope/3D-Speaker

Health v2 · maintenance only

Flags: no_releases

Adoption not part of the score

3121 stars · 268 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

3D-Speaker is an open-source Python toolkit for single- and multi-modal speaker verification, speaker recognition, and speaker diarization, with pretrained models hosted on ModelScope. It includes recipes and benchmarks for models like CAM++, ERes2Net, ECAPA-TDNN, and SDPN, plus a large-scale speech corpus for speech representation research.

Use cases

• verify speaker identity from voice recordings
• diarize multi-speaker meeting audio to see who spoke when
• extract speaker embeddings from audio
• train speaker verification models on VoxCeleb or CNCeleb
• identify spoken language from audio
• run multi-modal speaker recognition combining audio and video

When to choose

• you need state-of-the-art speaker verification or diarization with pretrained models
• you want reproducible recipes and benchmarks on VoxCeleb, CNCeleb, or 3D-Speaker datasets
• you work in Python/PyTorch and want ModelScope model integration

When to avoid

• you need general speech-to-text transcription rather than speaker analysis
• you need a production-ready plug-and-play service without training infrastructure
• you work outside Linux/PyTorch environments

Facets

library · maturity active

speech-recognition machine-learning audio-processing benchmarking sdk speech-processing machine-learning deep-learning python speaker-verification speaker-diarization speaker-recognition voice-embeddings pytorch pretrained-models language-identification voxceleb audio linux gpu

Member repositories

For agents

markdown · JSON · MCP: product_card(name="modelscope/3D-Speaker")