README.md

metadata

language:
  - hi
  - en
tags:
  - audio
  - text-to-speech
  - hinglish
task_categories:
  - text-to-speech
dataset_info:
  features:
    - name: audio
      dtype:
        audio:
          sampling_rate: 16000
    - name: text
      dtype: string
    - name: duration
      dtype: float64
    - name: source_index
      dtype: int64
  splits:
    - name: train
      num_bytes: 32321993
      num_examples: 500
  download_size: 32157293
  dataset_size: 32321993
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*

Hinglish Audio Dataset

This dataset contains 30 audio-text pairs.

Structure

• file_name: Audio file path
• text: Hinglish transcript
• duration: Duration in seconds

Audio was generated using Sarvam AI's Bulbul v2 model.