Skip to content

Dataset · speechcolab

ASR corpus for Thai, Indonesian, and Vietnamese

speechcolab/gigaspeech2

30k hours of auto-transcribed speech and a cleaned 22k-hour set for ASR in three languages.

What's inside

Large multilingual ASR corpus for low-resource languages: Thai, Indonesian, and Vietnamese. The raw part has about 30k hours of automatically transcribed speech; the refine part has 10k hours of Thai and 6k hours each of Indonesian and Vietnamese. This is audio and text for automatic-speech-recognition tasks.

What you can do with it

  • Fine-tuning an ASR model for Thai, Indonesian, or Vietnamese
  • Training a multilingual ASR system on low-resource languages
  • Evaluating speech recognition quality in three languages
Tasks
Automatic speech recognition
Languages
Thai, Indonesian, Vietnamese
Size
10M–100M rows
Downloads this month
11.6K
Likes
77
Access
on request on Hugging Face

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("speechcolab/gigaspeech2")