Logo Lanfrica

LKoech/MusicTranslateApp

Domaine:

natural language processing

Type de record:

software
Créateur:
LKo
Hôte:
Translating the music I love to english or any language I understand. I just love South African Music I just wanna know what it means through this app. # MusicTranslateApp A Gradio app that listens to spoken Zulu or Xhosa, transcribes it with Whisper, and is structured to translate the result into English. ## What it does - Records audio from your microphone in a web UI. - Runs speech-to-text with `openai/whisper-small` using Hugging Face Transformers. - Lets you choose source language: `Zulu` or `Xhosa`. - Returns recognized text live in the interface. ## Current status - Transcription path is active and working. - GMI-based translation integration exists in code (`translate_text_gmi`) but is currently commented out in the UI flow. ## Tech stack - Python - Gradio - PyTorch - Transformers (Whisper) - Librosa + NumPy - Requests ## Requirements - Python 3.10+ recommended - Optional NVIDIA GPU (falls back to CPU automatically) - Internet access to download model weights on first run ## Setup 1. Create and activate a virtual environment. 2. Install dependencies: ```bash pip install -r requirements.txt ``` ## Optional environment variable If you want to enable translation through the GMI API, set: ```bash GMI_API_KEY=your_api_key_here ``` PowerShell example: ```powershell $env:GMI_API_KEY="your_api_key_here" ``` ## Run ```bash python translator.py ``` Then open the Gradio URL shown in the terminal. ## How to use 1. Select `Zulu` or `Xhosa` in the dropdown. 2. Speak into your microphone. 3. Read the output text in the result box. ## Notes - First launch can take a while because Whisper model files are downloaded. - CPU inference can be slow; GPU significantly improves responsiveness. - The output label currently says "English Translation", but in the current code path it returns transcribed text. ## Project files - `translator.py`: App logic, Whisper pipeline, and Gradio UI. - `requirements.txt`: Python dependencies. ## Roadmap ideas - Re-enable translation in `voice_to_text`. - Add error handling and user-facing status messages. - Add language auto-detection and confidence display. - Add tests for prep …