Train on how people actually talk!
Conversational speech data in 60 languages and dialects, ready to license or recorded to your spec.
60Languages & dialects
73kHours of audio
7,250Native speakers
$60-95Per hour of audio
Meet the founder
The hard part isn’t the training
For custom speech data collection, we recruit native speakers, record to your brief, check each session and deliver audio, transcripts and consent documentation. Andreas walks through the collection process.
Featured datasets
Speech data your models are missing
Every dataset is spontaneous conversation, not read scripts: recorded in-country, dual-channel, with time-aligned transcripts and per-speaker metadata.
$90/hr
Spanish (Latin American)
Mexico · ConversationalHindi
India · ConversationalMandarin Chinese (Taiwanese)
Taiwan · ConversationalArabic (Modern Standard)
Saudi Arabia · ConversationalSwahili
Kenya · ConversationalWhy teams buy from us
Built for training, cleared for commercial use
Spontaneous, not scripted
Real two-person conversations with natural disfluencies, turn-taking, and code-switching. These are the acoustics your ASR model meets in production, not a reading booth.
Consent chain, documented
Every speaker signed an explicit commercial-use consent. We hand you the documentation, so GDPR and EU AI Act reviews don’t stall your procurement.
Transcripts & metadata included
Time-aligned, QA’d transcripts plus speaker age, gender, region, and recording-device metadata, ready for filtering, balancing, and evaluation splits.
Samples before you commit
Request free samples from any dataset and run your own audit: audio quality, transcript accuracy, speaker diversity. Judge the data, not the deck.
Transparent per-hour pricing
$60-$95 per hour depending on language, published on every dataset page. Volume and multi-language discounts on request.
Custom collection on tap
Need a language, dialect, or domain we don’t stock? Our custom collection network records to your exact spec.
Coverage
Eight regions, 60 languages and dialects
Europe (22)South Asia (10)Southeast Asia (6)East Asia (3)Middle East (7)Africa (8)North America (1)Latin America (3)
From high-demand markets like US English, German and Japanese to underserved languages like Swahili, Yoruba, Amharic and Moroccan Darija, including five Arabic variants and ten Indian languages.