README.md

---
language:
- zh
license: cc0-1.0

tags:
- audio
- speaker diarization

source_datasets:
- commonvoice

task_categories:
- automatic-speech-recognition

annotations_creators:
- crowdsourced

language_creators:
- crowdsourced
---
# cv-corpus-17.0-zh-TW-client_id-grouped

This dataset is a subset of the Common Voice dataset, filtered and grouped based on the client ID (treated as speaker ID).

## Dataset Details

- The dataset is derived from the Common Voice dataset.
- The original dataset is available at [Common Voice Dataset](https://commonvoice.mozilla.org/en/datasets).
- The dataset is grouped by client ID, which is treated as the speaker ID for this dataset.
- Each group is filtered to include only client IDs with a minimum of 30 samples and a maximum of 300 samples.
- The dataset is split into train and validation sets for each client ID group, with a ratio of 8:2.
- The same client IDs exist in both the train and validation sets.
- The dataset is split into batches of 1000 samples and saved as Parquet files.


## Dataset Statistics

- Filtered client_id count: 366
- Filtered total entry count: 45,229
- Original total entry count: 82,483

## Sample Duration Distribution

![Sample Duration Distribution](duration_distribution.png)

The histogram shows the distribution of sample durations in the dataset.

## License

The Common Voice dataset is licensed under the Creative Commons Zero (CC0) license.