Dataset · speechbrain
25,000 hours of English speech with transcripts
speechbrain/LoquaciousSet
A large heterogeneous English speech corpus with transcriptions and ready-made subsets for training ASR models.
What's inside
The dataset includes 25,000 hours of transcribed English speech—both read and spontaneous, recorded in clean and noisy conditions. The data is split into 6 subsets of varying size and quality (including large at 25,000 hours, medium at 2,500, small at 250, and clean at 13,000 hours) and is suitable for both research and commercial use.
What you can do with it
- Training and fine-tuning automatic speech recognition (ASR) models
- Testing speech recognition robustness against background noise and spontaneous speech
- Tasks
- Automatic speech recognition
- Languages
- English
- Size
- 10M–100M rows
- Downloads this month
- 25.7K
- Likes
- 64
Models trained on it
How to load it
from datasets import load_dataset
ds = load_dataset("speechbrain/LoquaciousSet")