Skip to content
The British Voices Bank
Open voice data

Open data for
the languages
the world forgot.

A UK-national archive of consensually sourced, linguistically verified voice data. Free for research, education, and public-interest AI.

Register interest → How it works →
CC BY 4.0 Consent-first Linguist verified
01 · The gap

AI speaks fewer than 100
of the world's 7,000 languages.

When a language can't be heard by the machines that mediate daily life, it slips further from the modern world. This is solvable.

0
Languages spoken
across Africa
< 0
Represented in
modern voice AI
0
World's languages in
open voice datasets
02 · How it works

A national asset, built from a company's commitment.

Every hour of commercial voice work we produce generates a deposit of verified, consented voice data for the public archive.

01

Consent-first

Contributors consent explicitly to how their data will be used. No scraping. No collection without consent.

Informed · specific · revocable
02

Linguist verified

Trained linguists review every batch. Transcriptions, dialect tags, and metadata are checked before archival.

Same standard as commercial
03

Deposited

A minimum of 20% of our annual commercial production is deposited into the Bank, by language family and dialect.

20% commitment · public pledge
04

Openly licensed

Released under CC BY 4.0. Available via API, bulk download, and curated collections. Free to approved users.

API · bulk · curated sets
03 · The archive

What's inside, and what's coming.

Launch languages, each with multiple dialects, genders, ages, and regional accents.

Nigerian Pidgin
Launch
Yoruba
Launch
Igbo
Launch
Hausa
Launch
Swahili
Launch
Amharic
Launch
Zulu
Launch
Xhosa
Launch
Twi
Launch
Wolof
Launch
Q2 2027
1,000 hrs
10 African languages
End 2027
10,000 hrs
25 languages + Commonwealth
2030
100,000+ hrs
100+ languages worldwide
04 · Who it's for

Free, for the people and projects that need it most.

🏛

UK Government & public bodies

Public-interest AI, educational tooling, digital-service accessibility.

No fees · full access
🔬

Academic researchers

Linguistics, AI, phonology, dialectology, model-fairness research.

Attribution requested · use freely
📚

Educators & schools

Language-learning tools, heritage language teaching, Saturday schools.

No fees · educational partnerships
⚡

Independent AI developers

Inclusive products with reasonable scale thresholds. Larger commercial use handled separately.

Free tier · commercial licence available
05 · Trust

Built on consent. Verified by linguists. Held in trust.

🔒

Consent-first

No voice data enters the Bank without explicit, informed consent. Withdraw any time. We honour it within 30 days.

✓

Linguistically verified

Trained linguists verify every recording for transcription accuracy, annotation, and language tagging before archival.

🌐

Open by default

All Bank data released under an open licence. Sensitive content gated to approved researchers with ethics review.

06 · Roadmap

Where we are, where we're going.

Q4 2026 · Now

Initiative announced

Partnership conversations begin. Interest registration opens.

Q1 2027

Governance board convened

First pilot deposits: Nigerian Pidgin and Yoruba, ~100 hours.

Q2 2027

Public launch

10 African languages, ~1,000 hours. API and bulk download opened.

Q4 2027

10,000 hours

25 African languages. First Commonwealth language deposits.

2028+

Commonwealth expansion

Caribbean, South Asian, Pacific languages. Educational partnerships at scale.

2030

100,000+ hours · 100+ languages

The world's largest consensually sourced, open voice archive.

07 · Get involved

Be part of the Bank from day one.

Register interest

Get notified when the Bank launches. Updates on deposits, API access, and partnership opportunities.

Register below ↓

Partner with us

Universities, government bodies, foundations: formal partnerships, dataset access, governance board seats.

Get in touch →

Contribute your voice

Native speakers of under-represented languages: get paid, help build the Bank, keep your language alive.

Start contributing →

Press & media

Covering AI inclusion, language preservation, or UK AI policy? We're here.

Press contact →

Register your interest

We'll notify you when the Bank launches and share updates on deposits and API access.

Contribute & licence

Everything you need to know.

What it means to contribute
You record short pieces of speech, conversation, and language improv, guided by prompts, in your native language. Every recording is reviewed by a linguist. Every contribution gets you paid. Every voice gets credited.
What you earn
We pay competitively for every verified minute of voice contributed. Rates vary by language and demand. Transparency is a core principle: rates are published on our contributor dashboard.
What happens to your voice
Your voice data is used in two ways: in Accent Studio's commercial dataset for AI labs, and in the British Voices Bank for public use. You consent to both uses explicitly. You can withdraw at any time.
Who can contribute
Native speakers of any under-represented language. Starting with African languages, expanding to Commonwealth and global languages over 2027 and beyond.
The licence (CC BY 4.0 target)
Bank data will be released under Creative Commons Attribution 4.0 International. Use it freely for research, education, and non-commercial AI development with credit to the Bank and its contributors. Large-scale commercial use is handled separately under a commercial licence.
Attribution
Cite the British Voices Bank, include the dataset version, and credit individual contributors where they have requested named attribution.
Governance
Maintained by Accent AI Studio Ltd. An external governance board (in formation) of linguists, data ethicists, Commonwealth cultural representatives, and public-interest technologists will review operational decisions before launch.
Compliance
Operated in compliance with UK GDPR, the Data Protection Act 2018, and relevant international frameworks. Full compliance statement will be published at launch.