Local speech-to-text with AMD ROCm GPU acceleration

A browser-based voice transcription app powered by OpenAI's Whisper model, optimized for AMD GPUs using ROCm. Built specifically for the new AMD Ryzen AI processors with Radeon 800M series integrated graphics.

• Real-time audio visualization with cyberpunk-themed UI
• Press-and-hold recording (mouse or touch)
• GPU-accelerated transcription via ROCm/CUDA
• One-click copy to clipboard
• Mobile-responsive design
• Automatic language detection

Tested on:

• AMD Ryzen AI 9 HX 370 + Radeon 890M (Strix Point / gfx1150)
• TUXEDO laptop running Ubuntu 24.04

Compatible with:

• AMD GPUs with ROCm support (gfx1150, gfx1100, etc.)
• NVIDIA GPUs (CUDA)
• CPU fallback (slower, but works)

git clone https://github.com/M64GitHub/whisper-rocm.git
cd whisper-rocm

python -m venv venv
source venv/bin/activate

For AMD Radeon 890M / 880M (gfx1150 - Strix Point):

pip install --index-url https://repo.amd.com/rocm/whl/gfx1150/ torch

For other AMD GPUs, check available builds at: https://repo.amd.com/rocm/whl/

For NVIDIA GPUs or CPU, see: https://pytorch.org/get-started/locally/

pip install -r requirements.txt

python App.py

Open http://localhost:8000 in your browser.

How to use:

1. Click and hold the "HOLD TO RECORD" button
2. Speak clearly into your microphone
3. Release the button to transcribe
4. Click "COPY" or use the auto-selected text

Edit App.py line 20 to change the model:

model = whisper.load_model("medium", device=device)  # Options: tiny, base, small, medium, large

• Backend: FastAPI + Uvicorn
• ML Model: OpenAI Whisper
• GPU Acceleration: PyTorch + ROCm 7.10
• Frontend: Vanilla HTML/CSS/JavaScript
• Audio: Web Audio API + MediaRecorder

source venv/bin/activate
python -c "import torch; print(torch.cuda.is_available()); print(torch.cuda.get_device_name(0) if torch.cuda.is_available() else 'No GPU')"

Make sure your browser has microphone access.

MIT License — see LICENSE for details.

• OpenAI Whisper — the speech recognition model
• AMD ROCm — GPU compute platform
• FastAPI — modern Python web framework