Check the separate README.md file which is in subfolder in tts_data_pipeline for more details

• Crawler Module: README.md
• Pre-Processing Module: README.md
• Alignment Module: README.md

A Python CLI tool for processing structured audio datasets, calculating total duration, and pushing the dataset to the Hugging Face Hub. Ideal for preparing Vietnamese TTS datasets or other speech-related corpora.

dataset/
├── speaker_001/
│   ├── audio1.wav
│   ├── audio1.txt
│   ├── audio2.wav
│   └── audio2.txt
├── speaker_002/
│   ├── ...

Each .wav file must have a corresponding .txt transcription file with the same name.

pip install typer librosa datasets num2words

python tts_data_pipeline/push_to_hub.py

Options:

• --input-dir: Path to the input dataset directory containing speaker subdirectories.
• --dataset-id: Dataset ID on Hugging Face Hub.
• --commit: Commit message for pushing to Hugging Face Hub.
• --private: Make the dataset private on Hugging Face Hub.
• --push-to-hub: Push the dataset to Hugging Face Hub.
• --calc-duration: Calculate total duration and print it to the console.

This project is licensed under the Apache License, Version 2.0.