README.md

metadata

dataset_info:
  features:
    - name: id
      dtype: string
    - name: src_audio
      dtype: audio
    - name: src_speaker
      dtype: string
    - name: src_text
      dtype: string
    - name: tgt_text
      struct:
        - name: cmn
          dtype: string
        - name: deu
          dtype: string
        - name: fra
          dtype: string
        - name: ita
          dtype: string
        - name: spa
          dtype: string
    - name: label
      dtype: string
  splits:
    - name: dev
      num_bytes: 443742514.64
      num_examples: 4790
    - name: test
      num_bytes: 614409814.469
      num_examples: 5757
  download_size: 1118172807
  dataset_size: 1058152329.109
configs:
  - config_name: default
    data_files:
      - split: dev
        path: data/dev-*
      - split: test
        path: data/test-*
license: cc-by-nc-4.0
task_categories:
  - translation
language:
  - es
  - it
  - fr
  - de
tags:
  - speech_translation

This is as unofficial mirror of the mExpresso dataset speech-to-text subset released under the Seamless model release by Meta. Original link can be found here https://github.com/facebookresearch/seamless_communication/tree/main/docs/expressive.