metadata
dataset_info:
features:
- name: audio
dtype: audio
- name: text
dtype: string
- name: start_time
dtype: string
- name: end_time
dtype: string
splits:
- name: train
num_bytes: 5723314
num_examples: 46
- name: validation
num_bytes: 1555591
num_examples: 16
download_size: 7274817
dataset_size: 7278905
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: validation
path: data/validation-*
dataset_description: >
This dataset contains transcriptions of audio segments, primarily designed for
fine-tuning automatic speech recognition (ASR) models like OpenAI's Whisper.
Each sample includes an audio clip, its corresponding transcription, and
timestamp information (start and end times) for the segment.
The data is organized into two splits:
- **Train:** 46 examples (~5.7MB)
- **Validation:** 16 examples (~1.5MB)
The dataset is particularly useful for tasks requiring timestamped speech
recognition.
usage: |
To load this dataset with the 🤗 Datasets library:
```python
from datasets import load_dataset
dataset = load_dataset("your-username/your-dataset-name")