A multilingual conversational-speech dataset with transcript alignment, speaker-turn and diarization information, language and accent metadata, and audio-quality signals.
Datoric’s ASR Voice Dataset is a collection of conversational speech intended for training and evaluating speech-recognition systems. The product page describes 48 kHz WAV audio, coverage across more than 30 languages, transcripts, and metadata for language, locale, accent, speakers, and audio quality. Speaker turns and diarization are available where specified for a subset. CSV and JSON metadata formats are described, with exact fields and delivery scope confirmed for the selected subset. Datoric says buyers can review sample materials and licensing documentation before selecting a delivery.
Be the first Cosign
Add it to your Products I use list. You can also share why you use it.
