Hugging Face
Models
Datasets
Spaces
Community
Docs
Enterprise
Pricing
Log In
Sign Up
Datasets:
ICTNLP
/
StreamUni
like
1
Follow
Natural Language Processing Group, Institute of Computing Technology, Chinese Academy of Science
96
Tasks:
Translation
Modalities:
Audio
Formats:
soundfolder
Languages:
Chinese
English
French
+ 2
Size:
1K - 10K
ArXiv:
arxiv:
2507.07803
Libraries:
Datasets
Croissant
License:
apache-2.0
Dataset card
Data Studio
Files
Files and versions
Community
1
Dataset Viewer
Auto-converted
to Parquet
API
Embed
Data Studio
Subset (1)
default
·
9.63k rows
default (9.63k rows)
Split (1)
train
·
9.63k rows
train (9.63k rows)
Search is not available for this dataset
SQL
Console
audio
audio
duration (s)
8.18
113
End of preview.
Expand
in
Data Studio
Previous
1
2
3
...
97
Next
Model
Github
The training dataset for the paper '
StreamUni: Achieving Streaming Speech Translation with a Unified Large Speech-Language Model
'
Model
https://huggingface.co/ICTNLP/StreamUni-Phi4
Github
https://github.com/ictnlp/StreamUni
Downloads last month
1,786
Use this dataset
Size of downloaded dataset files:
693 MB
Size of the auto-converted Parquet files:
684 MB
Number of rows:
9,631
Models trained or fine-tuned on
ICTNLP/StreamUni
ICTNLP/StreamUni-Phi4
Audio-Text-to-Text
•
6B
•
Updated
3 days ago
•
5