Meta FAIR Speech

by Meta AI / FAIR

Meta AI speech research — wav2vec 2.0 + HuBERT + MMS + Seamless origins.

Looking at Meta FAIR Speech? Try this first.

Drop your audio. Transcript in seconds. First transcript free, then $4 = 300 min

Upstream is archived. Meta FAIR Speech’s repository has been set read-only by its owner, so there are no further releases, fixes, or security patches. The last commit landed 2025-09-30. It can still be cloned and run, and forks may be maintained elsewhere. See the repository.

TL;DR

Best for self-supervised speech + multilingual ASR + speech translation at frontier scale. Pricing: free.

Category
Open source
License
MIT
Stars
★ 32.2k
Last push
2025-09-30
Pricing
free
Platforms
Web, GitHub

What it is

Meta FAIR Speech released wav2vec 2.0, HuBERT, data2vec, XLS-R, MMS (1100+ languages), Seamless, and SeamlessM4T. The industry frontier for open multilingual speech.

Best for: Self-supervised speech + multilingual ASR + speech translation at frontier scale.
Watch out for: MIT (fairseq) · per-model licenses vary (MMS = CC-BY-NC-4.0, Seamless = research) — check release.

Install / use

Features

Speaker diarizationNo
Word-level timestampsNo
Streaming / real-timeNo
Languages supported1100
HIPAA eligibleNo

Links

GitHub repo ↗

Meta FAIR Speech vs Whipscribe

FeatureMeta FAIR SpeechWhipscribe
CategoryOpen sourceTranscription APIs
Pricingfree$4–$24 one-time packs (credits never expire) · $2 single unlock · 60 min free on signup
Speaker diarizationNoYes
Word timestampsNoYes
StreamingNoNo
Languages110099
PlatformsWeb, GitHubWeb, API, MCP

Alternatives to Meta FAIR Speech

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.