Licensed Indic speech and text data for AI, from consented contributors
4.6(2)4 published listings.
38,000 hours of transcribed two-speaker conversations in 22 Indian languages from consented contributors, with dialect, age band, environment and device metadata.
Speech recorded to your script, languages, demographics and acoustic conditions by consented contributors in India and South Asia, transcribed and delivered with consent records.
Native-speaker annotators for RLHF preference ranking, response rating, classification and entity tagging in 22 Indian languages, with gold checks and agreement reports.
62 billion deduplicated tokens of licensed text in 22 Indian languages and Indian English, with source, licence, domain and quality tags on every document.
Kathana Language Data builds speech and text datasets for Indian languages with a network of more than 40,000 consented contributors and in-house linguists in Pune and Hyderabad. AI teams license its corpora and commission annotation, RLHF and custom collection projects for ASR, TTS and language models.
What the provider licenses its data for, and what it does not allow.
Training and evaluating speech recognition, speech synthesis and language models; voice assistants and contact-centre analytics.
Speaker identification or verification, cloning the voice of an individual contributor, biometric profiling, and any attempt to re-identify contributors.
DataStoa's acceptable use policy applies on top: no re-identification, discrimination, surveillance or sanctioned uses.
Provider-reported. Ask for certificates or a data processing agreement before you sign.
Other places you can receive this provider's data, useful if your team already buys through a cloud marketplace.
Buyers rate four things: data quality, volume and coverage, value for money and service.
from 2 reviews
The 2,000-item preference pilot came back on time with agreement metrics and annotator notes in Hindi and English. Guideline changes were absorbed within a day. We have since scaled to several thousand comparisons a week.
Meera J., Applied scientist, Enterprise AI lab
Every hour of audio links to a consent record that covers commercial model training, and the speaker metadata let us balance dialects properly. Transcription quality in Tamil and Bengali was the best we have bought.
Vikram R., Speech ML lead, Voice assistant company