You need to agree to share your contact information to access this dataset

This repository is publicly accessible, but you have to accept the conditions to access its files and content.

Log in or Sign Up to review the conditions and access this dataset content.

Kom Audio-Text Dataset

Kom text and audio pairs prepared for speech research.

Source and attribution

The source reference is provided for provenance. Users must confirm and follow the source provider's licensing and redistribution terms.

Fields

  • audio: WAV audio resolved by the AudioFolder loader
  • text: normalized Kom chapter text
  • duration_seconds, sampling_rate, channels: audio properties

Important

These examples are complete chapters. Segment and align them before ASR or TTS training. Confirm and document redistribution rights for the source audio and text.

Downloads last month
15