Speech datasets obtained from the Living Audio Datasets project:

• Irish

  • Source: audio, transcripts

  • Original data: audio (273MB), transcripts

  • Coqui STT (95MB)

  • Festvox (95MB)

  • The convert.sh scripts converts the original data into Festvox and Coqui STT (uses wai.annotations for the latter)

License

Apache 2.0