README.md

metadata

dataset_info:
  features:
    - name: audio
      dtype: audio
    - name: transcription
      dtype: large_string
    - name: speaker_id
      dtype: large_string
    - name: sentence_id
      dtype: large_string
    - name: sentence_domain
      dtype: large_string
    - name: up_votes
      dtype: int64
    - name: down_votes
      dtype: int64
    - name: age
      dtype: large_string
    - name: gender
      dtype: large_string
    - name: accents
      dtype: large_string
    - name: variant
      dtype: large_string
    - name: locale
      dtype: large_string
    - name: segment
      dtype: large_string
    - name: split
      dtype: large_string
  splits:
    - name: train
      num_bytes: 1033871268
      num_examples: 25198
  download_size: 930519177
  dataset_size: 1033871268
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*

common-voice-scripted-speech-quebec

This dataset is a filtered subset of the Mozilla Common Voice Scripted Speech 25.0 - French dataset. It exclusively contains audio clips from speakers with Canadian and Québécois accents.

Dataset Summary

• Language: French (fr)
• Total Clips: 25,198
• Total Duration: 36.83 hours (132,578.64 seconds)
• License: CC-0

Filtering Criteria

This subset was generated by extracting rows from the original cv-corpus-25.0-2026-03-09 dataset that match the following criteria:

• variant is exactly "Français d'Amérique du Nord"
• OR accents matches any self-reported Canadian/Québécois identifier (e.g., "Français du Canada", "Québécois", "Français du Québec|Saguenay").

Data Structure

The dataset is formatted as a Hugging Face Dataset with the audio cast to the Audio() feature type.

Feature	Type	Description
audio	Audio	The loaded audio dictionary (bytes/path) and sampling rate
transcription	string	The text transcription of the audio
speaker_id	string	Hashed UUID of the contributor
sentence_id	string	Unique identifier for the text prompt
sentence_domain	string	Domain(s) the sentence belongs to
up_votes	int64	Number of users who validated the clip
down_votes	int64	Number of users who rejected the clip
age	string	Self-declared age bracket
gender	string	Self-declared gender
accents	string	Self-declared accent
variant	string	Self-declared language variant
locale	string	Original locale code (fr)
segment	string	Custom dataset segment identifier
split	string	Original dataset split (e.g., train, dev, test, validated)

Source and Documentation

For comprehensive information regarding the original data collection methodology, text corpus sources, demographic definitions, and broader project details, please refer to the official datasheet:

https://mozilladatacollective.com/datasets/cmn5zugst00w3nv07upovf2bg