Dataset · nvidia
NVIDIA multilingual audio instruction dataset
nvidia/Granary-v2
Generated labels and instruction data for multilingual speech, covering ASR, translation, and speech question answering.
What's inside
The second release of NVIDIA's audio dataset featuring refined annotations across a wide range of source speech corpora. Along with transcriptions and translations, it includes labels for text normalization, audio captioning, code-switching, and speech QA pairs. Data is available in multiple languages including Arabic, Assamese, Bulgarian, Bengali, Czech, and Danish.
What you can do with it
- Training and fine-tuning multilingual automatic speech recognition (ASR) and machine translation models
- Training multimodal audio-text models for speech question answering (speech-QA)
- Adapting models for code-switching and text normalization
- Tasks
- Automatic speech recognition, Translation, Audio text → text, Question answering
- Languages
- Arabic, Assamese, Bulgarian, Bangla, Czech, Danish, German, Greek and 4 more
- Downloads this month
- 10.6K
- Likes
- 61
Models trained on it
- nvidia/NVIDIA-Nemotron-3.5-Transcribe-30B-A4BAutomatic speech recognition 94.5K
- nvidia/NVIDIA-Nemotron-3.5-Transcribe-30B-A4B-FP8Automatic speech recognition 73.9K
- nvidia/NVIDIA-Nemotron-3.5-Transcribe-30B-A4B-NVFP4Automatic speech recognition 23.4K
- prithivMLmods/NVIDIA-Nemotron-3.5-Transcribe-30B-A4B-GGUFAutomatic speech recognition 4.8K
- AMAImedia/NVIDIA-Nemotron-3.5-Transcribe-30B-A4B-BF16-GGUFAutomatic speech recognition 1.6K
- mradermacher/NVIDIA-Nemotron-3.5-Transcribe-30B-A4B-i1-GGUF 1.5K
- mradermacher/NVIDIA-Nemotron-3.5-Transcribe-30B-A4B-GGUF 1.5K
- KirillFPV/nemotron-transcribe-partsAutomatic speech recognition 331
How to load it
from datasets import load_dataset
ds = load_dataset("nvidia/Granary-v2")