The dataset is currently empty. Upload or create new data files. Then, you will be able to explore them in the Dataset Viewer.

AtlasSpeech Dialogues

AtlasSpeech Dialogues contains consented conversational audio segments and aligned transcripts for robustness research.

Data fields

Each record contains an audio reference, transcript, speaker split, and recording environment tag.

Access notes

Users should retain the supplied split identifiers when reporting benchmark results.

Documentation stewardship

  • Dataset: toolathlonEval/AtlasSpeech-Dialogues
  • Standard: Open Dataset Card Guide 2.1
  • Review cadence: Annual
  • Support channel: canonical GitHub issue
Downloads last month
10