Welcome to Voxdu, an elegantly engineered desktop application dedicated to offline Urdu speech recognition. Designed with a focus on both aesthetic brilliance and computational efficiency, Voxdu transforms spoken Urdu into highly accurate text using state of the art artificial intelligence.
Voxdu is constructed upon a robust foundation, integrating several advanced technologies to deliver a seamless user experience:
- Core Inference Engine: Utilizes
faster-whisperfor highly optimized, offline transcription, ensuring your data remains completely private and secure. - Audio Pipeline: Features sophisticated audio processing capabilities, including real time noise reduction, bandpass filtering, and audio extraction from video files via FFmpeg.
- User Interface: Crafted with PyQt6, the interface offers a deeply immersive experience complete with fluid animations, dynamic waveforms, and multiple bespoke themes (Dark, Light, AMOLED, Sepia, Slate).
- Intelligent Resource Management: The application employs multithreading for all resource intensive tasks, guaranteeing a responsive UI even during heavy transcription workloads.
- Completely Offline Operation: Once the initial model is downloaded, Voxdu requires zero internet connectivity.
- Microphone & File Support: Transcribe live audio through your microphone or import existing audio and video files.
- Advanced Noise Reduction: Built in algorithms clean up noisy audio sources before passing them to the inference engine, drastically improving accuracy.
- Rich Export Options: Save your transcriptions as plain text, subtitle files (SRT), or formatted Microsoft Word documents (DOCX).
- Customizable AI Parameters: Fine tune the transcription process with adjustable beam sizes, temperature controls, and voice activity detection (VAD) settings.
Setting up Voxdu is designed to be frictionless.
- Prerequisites: Ensure you have Python 3.11 installed on your system. During installation, it is imperative to check the "Add Python to PATH" option.
- Automated Setup: Locate and execute the
setup.batfile. This intelligent script will automatically audit your environment, install FFmpeg (if absent), and provision all required Python dependencies in a carefully orchestrated sequence. - Launch: Once the setup concludes successfully, simply run
run.pyor double click it to launch the application.
Note: Upon your first transcription request, Voxdu will automatically download the required AI model to your system cache. Subsequent usages will be entirely offline.
- Language: Python 3.11+
- GUI Framework: PyQt6
- AI Backend: faster-whisper, CTranslate2
- Audio Processing: numpy, scipy, noisereduce, pyaudio, soundfile
- Media Handling: FFmpeg
Experience the intersection of beautiful design and powerful machine learning with Voxdu.