Dataset Viewer
The dataset could not be loaded because the splits use different data file formats, which is not supported. Read more about the splits configuration. Click for more details.
Couldn't infer the same data file format for all splits. Got {NamedSplit('train'): ('json', {}), NamedSplit('validation'): ('webdataset', {}), NamedSplit('test'): ('audiofolder', {})}
Error code:   FileFormatMismatchBetweenSplitsError

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.

YAML Metadata Warning:empty or missing yaml metadata in repo card

Check out the documentation for more information.

Causality-Aware CVSS Data Resources

This repository contains generated resources for building causality-aware speech-to-speech translation (S2ST) data from CVSS.

The resources are designed for causality-aware adaptive policy experiments FAST-CAP.

This repository is derived from the CVSS dataset, which is a multilingual-to-English speech-to-speech translation corpus built from Common Voice and CoVoST 2. This repository focuses on the following language directions:

  • Spanish → English
  • French → English
  • German → English

The repository includes examples from the train, dev, and test subsets.


Repository Structure

Causality-Aware-CVSS/
├── raw_sample/
├── generated_resources/
│   ├── a2flow_tts/
│   ├── awesome_align_clean/
│   └── mfa/
│       ├── cvss/
│       └── a2flow_tts/
│           └── cvss/
├── cap_prep_sample/
│   └── cvss_dev_es/
└── README.md

Directory Contents

  • raw_sample/: Source samples from the Spanish Dev CVSS-T dataset.

  • generated_resources/: Data resources generated by the causality-aware pipeline:

    • a2flow_tts/: Generated target speech with improved speaker similarity and preservation compared to original CVSS-T
    • awesome_align_clean/: Text-to-text alignments between source and target languages
    • mfa/: Speech-to-text alignments via Montreal Forced Aligner for both original and generated speech
  • cap_prep_sample/: Causality-aware samples in Lhotse shard format ready for training/validation

Citation

If you use this dataset in your research, please cite the paper below.

@inproceedings{
  
}

---
license: mit
Downloads last month
14