---
license: mit
task_categories:
- automatic-speech-recognition
- text-to-speech
tags:
- speech
- audio
---

# Audio Dataset

This dataset contains audio segments with transcriptions for speech recognition and text-to-speech tasks.

## Dataset Structure

- `dataset.json`: Complete dataset in JSON format
- `train.jsonl`: Dataset in JSONL format for use with Hugging Face datasets library
- `dataset_info.json`: Metadata about the dataset structure

Each entry contains:
- `file`: Original filename
- `audio`: Audio data with sampling rate
- `transcription`: Text transcription
- `duration`: Audio duration in seconds