CoolFace
Datasetpublic

yakhyo/mozilla-common-voice-uzbek

🗣️ Mozilla Common Voice (Uzbek) — Cleaned & Normalized This dataset is a refined version of mozilla-foundation/common_voice_17_0, containing only Uzbek language voice recordings, and enriched with preprocessing steps for better usability in training ASR models. 🔍 Dataset Overview This version focuses exclusively on Uzbek audio samples and includes the following modifications: 🎯 Filtered to include only Uzbek examples. ✨ Normalized text field added under the… See the full description on the dataset page: https://huggingface.co/datasets/yakhyo/mozilla-common-voice-uzbek.

sourceHugging Faceupdated 1y agoView on Hugging Face
4likes193downloads
Dataset Card

🗣️ Mozilla Common Voice (Uzbek) — Cleaned & Normalized

This dataset is a refined version of `mozilla-foundation/common_voice_17_0`, containing only Uzbek language voice recordings, and enriched with preprocessing steps for better usability in training ASR models.

🔍 Dataset Overview

This version focuses exclusively on Uzbek audio samples and includes the following modifications:

  • —🎯 Filtered to include only Uzbek examples.
  • —✨ Normalized text field added under the key text.
  • —✅ Aligned audio format and sampling rate to standard expectations (16kHz, mono).

🛠️ Modifications Applied

The original sentence field was normalized to remove inconsistencies in apostrophe usage. A new text column was added with normalized transcription:

python
def normalize_text(batch):
    batch["text"] = batch["sentence"].replace("‘", "'").replace("’", "'").replace("`", "'")
    return batch

📦 Usage

python
from datasets import load_dataset

dataset = load_dataset("yakhyo/mozilla-common-voice-uzbek")

# Access normalized text
print(dataset["train"][0]["text"])

📜 License

This dataset inherits the Mozilla Common Voice license (CC-0 1.0).

🙌 Acknowledgements