Dataset · litagin
Large Japanese speech corpus from visual novels
litagin/Galgame_Speech_ASR_16kHz
From 1 to 10 million pairs of 16 kHz audio and Japanese transcriptions for ASR tasks.
What's inside
The dataset contains 1 to 10 million Japanese audio-text examples sampled at 16 kHz, gathered from visual novel (galgame) voice acting. Distributed strictly for non-commercial use under the GPL v3.0 license, requiring open-sourcing the code of trained models.
What you can do with it
- Training or fine-tuning ASR models to recognize emotional Japanese speech
- Developing automatic captioning systems for Japanese games and anime in non-commercial projects
- Tasks
- Automatic speech recognition
- Languages
- Japanese
- Size
- 1M–10M rows
- Downloads this month
- 470
- Likes
- 48
Models trained on it
- litagin/anime-whisperAutomatic speech recognition 30.4K
- jaykwok/Qwen3-ASR-1.7B-JA-Anime-GalgameAutomatic speech recognition 4.6K
- quantumcookie/anime-whisper-ct2-int8Automatic speech recognition 1.5K
- mradermacher/Qwen3-ASR-1.7B-JA-Anime-Galgame-GGUFAutomatic speech recognition 741
- quantumcookie/anime-whisper-ct2-fp16Automatic speech recognition 622
- mradermacher/Qwen3-ASR-0.6B-JA-Anime-Galgame-GGUFAutomatic speech recognition 517
- AkitoP/whisper-large-v3-japense-phone_accentAutomatic speech recognition 233
- quantumcookie/anime-whisper-ct2Automatic speech recognition 94
How to load it
from datasets import load_dataset
ds = load_dataset("litagin/Galgame_Speech_ASR_16kHz")