Search is not available for this dataset
audio
audioduration (s)
4.8
15.7

This dataset can be used to fine-tune Speech To Text models as Text To Speech.

dataset information

  • Speaker: Aldo
  • Dataset size: 535 audio files
  • audio duration of 4-15 seconds (1:33:15)

Dataset structure

This dataset has been structured in the LJSpeech format:

  • wavs/
    • 1.wav
    • 2.wav
    • 3.wav

    • 535.wav
  • transcript.csv
Downloads last month
49
Edit dataset card

Models trained or fine-tuned on rmcpantoja/Ald_Mexican_Spanish_speech_dataset