Skip to content

Latest commit

 

History

History
49 lines (28 loc) · 1.36 KB

arm.md

File metadata and controls

49 lines (28 loc) · 1.36 KB
Raw

Development on Arm dev machine

Install Whisper.cpp

First, compile from the source code:

  1. git clone https://github.com/ggerganov/whisper.cpp.git
  2. sudo apt update && sudo apt upgrade && sudo apt install make gcc g++
  3. cd whisper.cpp
  4. make

Then, download models:

  • bash ./models/download-ggml-model.sh base-q5_1
  • ``

Test:

./main -m /home/ubuntu/whisper.cpp/models/ggml-base-q5_1.bin -f samples/jfk.wav

Install edge-whisper-server systemd service

copy edge-whisper-server.service to /etc/systemd/system/. Then, mkdir -p /home/ubuntu/edge-whisper, copy whisper_server.py and .env to this directory.

Install pip and ffmpeg:

sudo apt install python3-pip ffmpeg

Install pip dependencies:

pip install aiofiles uvicorn fastapi pydub openai-whisper python-multipart eyed3 scipy

Start this systemd service to run the backend services:

  1. sudo systemctl enable edge-whisper-server
  2. sudo systemctl start edge-whisper-server

Or, you can run dev server by:

ALLOW_ORIGIN=https://edge-ai.yomo.run WHISPERCPP_BIN=/home/ubuntu/whisper.cpp/main WHISPERCPP_MODEL=/home/ubuntu/whisper.cpp/models/ggml-base-q5_1.bin python whisper_server.py

Web App

Clone this repo and run pnpm install && pnpm run dev

Notice

You have to change the CORS settings following your dev environment.