VoxConverse

by Oxford VGG

50h audio-visual diarization corpus — wild YouTube speakers in conversation.

TL;DR

50h audio-visual diarization corpus — wild YouTube speakers in conversation.

Best for diarization benchmarks with realistic in-the-wild audio. Pricing: research-only.

Category
Open source
License
—
Stars
—
Last push
—
Pricing
research-only
Platforms
Web, GitHub

What it is

VoxConverse provides ~50h of audio-visual diarization data with speaker overlap, music, and crowd noise — harder than AMI. License: CC BY 4.0 metadata.

Best for: Diarization benchmarks with realistic in-the-wild audio.
Watch out for: CC BY 4.0 metadata · YouTube-sourced · variable speaker count per recording (1–20+). Cite: Chung et al., Interspeech 2020.

Install / use

git clone https://github.com/joonson/voxconverse

Features

Speaker diarizationYes
Word-level timestampsNo
Streaming / real-timeNo
Languages supported1
HIPAA eligibleNo

VoxConverse vs Whipscribe

FeatureVoxConverseWhipscribe
CategoryOpen sourceTranscription APIs
Pricingresearch-onlyfree beta
Speaker diarizationYesYes
Word timestampsNoYes
StreamingNoNo
Languages199
PlatformsWeb, GitHubWeb, API, MCP

Alternatives to VoxConverse

OpenAI Whisper
OpenAI

The reference open-source multilingual ASR model from OpenAI.

OSS · MIT ★ 98.1k
whisper.cpp
Georgi Gerganov

C/C++ port of Whisper — runs on anything, from a Raspberry Pi to Apple Silicon.

OSS · MIT ★ 48.8k
faster-whisper
SYSTRAN

4× faster than reference Whisper using CTranslate2 — production sweet spot.

OSS · MIT ★ 22.3k

Whipscribe is a managed faster-whisper + whisperX service. If you want transcripts without running infrastructure, paste a URL or drop a file in the form below — you'll have a transcript in seconds.