The dataset viewer is not available for this split.
Server error while post-processing the rows. Please report the issue.
Error code: RowsPostProcessingError
Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.
DuraSet-440K
English–Chinese speech-to-speech translation data for DuraS2ST, with paired audio, transcriptions, translations, and chain-of-thought annotations.
Dataset
| Subset | Direction | Records |
|---|---|---|
en-zh |
English → Chinese | 314,229 |
zh-en |
Chinese → English | 243,129 |
| Total | 557,358 |
The two subsets share audio stored in ZIP shards under audio/en/ and audio/zh/.
Metadata is stored in compressed Parquet files under data/<subset>/.
Usage
from datasets import load_dataset
ds = load_dataset("Mia11939/DuraSet-440K", "en-zh", split="train", streaming=True)
example = next(iter(ds))
Use "zh-en" for Chinese-to-English translation.
Fields
| Field | Description |
|---|---|
id |
Record identifier within each subset |
source_lang, target_lang |
Language codes: en or zh |
source_audio, target_audio |
Audio in Hugging Face Audio format |
source_text, target_text |
Source transcription and target translation |
source_duration, target_duration |
Audio duration in seconds |
cot |
Chain-of-thought annotation |
License
- Downloads last month
- 460