Have voice recordings? License them to AI teams through Vocalstock.
Datailix lists your existing speech data next to the recordings our speakers make, and licenses it to companies building speech and language AI. You keep your data and earn a fixed amount for every hour of it that a buyer licenses — every time it sells.
1. Request an accountIndividuals and companies are welcome. Tell us what data you have and how the speakers gave their consent; we review every request.
2. Upload a batchAt least 10 hours of one language, one domain and one recording set-up, with a CSV describing every file and its speakers.
3. We check itFormats, duplicates and a sample listened to by our team. Approved batches go on sale as off-the-shelf data.
4. Get paid per hour soldA fixed rate per hour, agreed for each batch, every time the data is licensed. You follow every sale on your page.
What we need
- Your rights: you own the recordings, and every speaker agreed in writing that their voice may be licensed to companies for AI training. We ask for your consent form and a few signed examples with each batch.
- Adults only: every speaker was 18 or older.
- Quality: WAV or FLAC, 16 kHz or more, original recordings (not converted from MP3 or similar).
- Speaker details: for every file, an id per speaker plus gender, year of birth and country. Buyers only ever see anonymous ids, gender, age range and country.
- Variety: at most 25 hours from any one speaker (or pair of speakers).
- Honesty: tell us if the same data is also sold elsewhere. Every file is fingerprinted, so the same audio can't be sold twice on Vocalstock.