🗣️ Persian Speech Recognition

A Persian (Farsi) speech recognition system designed to convert spoken Persian language into text. This project leverages deep learning models to achieve high-accuracy transcription for Persian speech.

🪄 Features

• Pre-trained models for Persian speech recognition
• Dataset support for training custom models

🛠️ Prerequisites

• Python 3.7+
• pip
• (Optional) CUDA for GPU acceleration

⚙️ 2 Environment Setup

1. Clone the repository:

   git clone https://github.com/IranTechNest/PersianSpeechRecognition.git 
   

   then

   cd PersianSpeechRecognition
   

2. Install dependencies:

   pip install -r requirements.txt 
   

🚀 Usage

1. Run app:

 python app.py

2. Speech Recognition You can choose from the following supported models for speech recognition:

• small
• small-v2
• medium
• large

from src.voice2txt import SpeechRecognition

speech = SpeechRecognition(model='medium')
pred_transcription = speech.speech_recognition("output_audio.wav") 

pred_transcription = [سلام من اومدم یک دو سه چهار]

🤝 Contributing

We love contributions from the community! Here's how you can help improve PersianSpeechRecognition:

• 🐛 Report bugs by opening an issue
• 💡 Suggest new features through GitHub issues
• 📝 Improve documentation (fix typos, add examples)
• 💻 Contribute code - see our Contribution Guide

First time contributing? Look for issues labeled [good first issue](https://github.com/IranTechNest/PersianSpeechRecognition/issues?q=is%3Aopen+is%3Aissue+label%3A%22good+first+issue%22

🙏 Acknowledgments

Thanks to Mozilla Common Voice for open-source datasets

Thank of https://huggingface.co/masoudkaviani for funtuing the models

Inspired by DeepSpeech and HuggingFace Transformers

📬 Contact

For questions or support, please open an issue or contact the maintainers at IranTechNest.

ℹ️ About PersianSpeechRecognition

A Persian speech recognition toolkit for developers and researchers...

📌 Citation

@misc{raminram_persianspeech,
  title        = {PersianSpeechRecognition},
  author       = {Ramin Rahimi},
  howpublished = {\url{https://github.com/IranTechNest/PersianSpeechRecognition}},
  year         = {2025},
  note         = {Accessed: 2025-06-02},
  email        = {raminram6970@gmail.com}
}