Skip to content

Dataset · anquachdev

English-Vietnamese speech translation corpus (Hibiki-Zero)

anquachdev/en-vi-speech-20k

Between 1M and 10M records for training simultaneous speech translation from Vietnamese to English, ASR, and TTS.

What's inside

The dataset is designed for fine-tuning the Hibiki-Zero simultaneous speech translation model in the Vietnamese-to-English direction. It covers translation, ASR, and TTS tasks, and one of the folders (default) contains 696,243 text pairs from the PhoMT corpus. File access is restricted but automatically approved after accepting the source license terms.

What you can do with it

  • Fine-tuning simultaneous speech translation models (Speech-to-Speech / Speech-to-Text)
  • Training ASR and TTS systems for Vietnamese and English languages
Tasks
Translation, Automatic speech recognition, Text → speech
Languages
English, Vietnamese
Size
1M–10M rows
Downloads this month
37.9K
Likes
0
Access
on request on Hugging Face

How to load it

from datasets import load_dataset

ds = load_dataset("anquachdev/en-vi-speech-20k")