VoxPopuli

by Meta AI / Facebook AI Research

400k hours of European Parliament speeches in 23 EU languages.

TL;DR

400k hours of European Parliament speeches in 23 EU languages.

Best for multilingual self-supervised pretraining; cross-lingual ASR; speech translation (15 source × 15 target). Pricing: free.

Category
Open source
License
—
Stars
—
Last push
—
Pricing
free
Platforms
HuggingFace, GitHub

What it is

VoxPopuli is 400k unlabeled + 1.8k transcribed + 17.3k interpreted hours of EU Parliament plenary recordings, 2009–2020. The de-facto multilingual pretraining corpus for European languages. License: CC0.

Best for: Multilingual self-supervised pretraining; cross-lingual ASR; speech translation (15 source × 15 target).
Watch out for: CC0 1.0 · domain-specific (parliamentary speech) · formal register · varying per-language coverage. Cite: Wang et al., ACL 2021.

Install / use

from datasets import load_dataset; ds = load_dataset('facebook/voxpopuli', 'en')

Features

Speaker diarizationNo
Word-level timestampsNo
Streaming / real-timeNo
Languages supported23
HIPAA eligibleNo

VoxPopuli vs Whipscribe

FeatureVoxPopuliWhipscribe
CategoryOpen sourceTranscription APIs
Pricingfreefree beta
Speaker diarizationNoYes
Word timestampsNoYes
StreamingNoNo
Languages2399
PlatformsHuggingFace, GitHubWeb, API, MCP

Alternatives to VoxPopuli

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.