Speech & audio

Collections

• Conversations in many languages

  750K+ hours

  Casual chats, interviews and debates in dozens of languages, with code-switching and dialects left in.

  Best for: Multilingual speech recognition, translation, language ID

  code-switching / dialects

• Everyday conversation

  500K+ hours

  Real calls and face-to-face talk, with the ums, interruptions and laughter left in.

  Best for: Robust speech recognition, voice assistants

  overlap / natural

• Sports commentary

  1M+ hours

  Play-by-play, analysis and crowd noise from one, two and three-person booths.

  Best for: Live captioning, event detection from audio

  live

• Studio interviews

  400K+ hours

  Studio-quality Q&As, podcast sit-downs, panels and testimonials.

  Best for: Clean speech recognition, speaker separation

  studio

• Podcasts with several hosts

  300K+ hours

  Full episodes with several speakers, crosstalk and all.

  Best for: Speaker separation, long-form transcription

  full episodes

• Scripted studio voice

  3K+ hours

  Voice-over, narration, audiobooks and e-learning recorded in treated rooms, up to 96 kHz.

  Best for: Text to speech, voice cloning, prosody

  studio / hi-res

• Sound effects, one sound at a time

  20K+ hours

  Foley, impacts, vehicles, UI sounds and ambience, each recorded on its own.

  Best for: Text to audio, sound for video, tagging

  isolated / up to 192 kHz

• Bengali, Hindi, Tamil and Telugu speech

  Volume on request

  Read and conversational speech with checked transcripts and speaker details.

  Best for: Indic speech recognition, speaker separation

  transcripts / speaker details

• Read speech in many languages

  750K+ recordings

  People reading prompts aloud in major world languages, lossless audio with transcripts.

  Best for: Multilingual speech recognition and testing

  many languages / lossless

• Regional accents and dialects

  Volume on request

  Mandarin, Cantonese, Henan, Urdu, Korean, Spanish and French, with native transcripts.

  Best for: Dialect and low-resource speech recognition

  dialects