MahmoudAshraf97/whisper-diarization

Health v2 · maintenance only

Flags: no_releases

Adoption not part of the score

5630 stars · 503 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-29, confidence not recorded

A pipeline that combines OpenAI Whisper transcription with speaker diarization using Voice Activity Detection (MarbleNet) and speaker embeddings (TitaNet). It aligns timestamps with ctc-forced-aligner to attribute each word in the transcript to the correct speaker.

Use cases

• transcribe a meeting recording with speaker labels
• add speaker diarization to whisper transcripts
• convert podcast audio into a who-said-what transcript
• identify speakers in an interview audio file
• transcribe multi-speaker audio with timestamps
• run speech-to-text with diarization on GPU

When to choose

• you need transcripts with per-word or per-sentence speaker attribution
• you want a ready-made pipeline combining Whisper, VAD, and speaker embeddings
• you want a Colab-friendly or CLI transcription tool with diarization

When to avoid

• you only need plain transcription without speaker labels
• you need real-time streaming diarization
• you cannot run GPU-accelerated models locally

Facets

library · maturity active

speech-recognition audio-processing machine-learning cli speech-processing artificial-intelligence python cross-platform whisper speaker-diarization asr transcription vad nemo speech-to-text natural-language-processing gpu

Member repositories

For agents

markdown · JSON · MCP: product_card(name="MahmoudAshraf97/whisper-diarization")