alakxender/voice-synthetic

Voice Synthetic Dataset

A Dhivehi text-to-speech dataset containing professionally recorded audio samples with their corresponding transcriptions.

Dataset Configuration

• Audio Format: WAV files
• Language: Dhivehi (dv)
• Source: cv
• Split Ratio:
  • Train: 80%
  • Test: 20%
• Structure:
  • audio: Path to the audio file
  • sentence: Corresponding Dhivehi text transcription
  • source: Source of the audio data

Data Format

{
    'audio': {
        'path': 'path/to/audio.wav',
        'array': array(...),
        'sampling_rate': 22050
    },
    'sentence': 'ދިވެހި ޓެކްސްޓް'  # Dhivehi text
    'source': 'cv'  # Source identifier
}