PawanKrd/tts-ckb

Dataset Summary

This dataset is designed for Text-to-Speech (TTS) applications in Central Kurdish (ckb), providing audio samples and corresponding transcriptions in the Central Kurdish language. It is useful for training and evaluating TTS models tailored for this specific language. The audio files are sampled at 16kHz, ensuring high-quality audio for speech synthesis tasks.

Dataset Structure

Features

• client_id (string): Unique identifier for each client (speaker).
• audio (audio): Audio data sampled at 16,000 Hz.
• sentence (string): Transcription of the audio data in Central Kurdish.

Splits

The dataset is divided into two main splits:

• Train
  • Number of Examples: 14,669
  • Total Size: 389.51 MB
• Test
  • Number of Examples: 1,630
  • Total Size: 43.28 MB

Dataset Size

• Download Size: 426.92 MB
• Dataset Size: 432.79 MB

Configuration

• Configuration Name: default
• Data Files:
  • Train Split: data/train-*
  • Test Split: data/test-*

Usage

This dataset can be used for training and evaluating TTS models in Central Kurdish. The audio files and their corresponding transcriptions can be fed into TTS models to generate synthetic speech in this language. The dataset’s structure and quality make it suitable for research and development in speech synthesis.

Additional Information

• Language: Central Kurdish (ckb)
• Sampling Rate: The audio files are sampled at 16,000 Hz.
• Total Number of Examples: 16,299