Perroquet

Discord voice message transcription bot using OpenAI's Whisper models

Installation / Usage

Clone this repo

Create a models directory at the root and download a binary from here.

Copy the .env.example into .env and fill it with your info.

Follow this guide to create a discord bot and add it to your server with this guide.

Docker (recommended)

docker build -t perroquet .
docker run -v ./.env:/app/perroquet/.env -v ./models/:/app/perroquet/models perroquet

From source

Clone the smart-whisper repository and compile libwhisper.a:

git clone --recursive https://github.com/JacobLinCool/smart-whisper.git
cd smart-whisper/whisper.cpp
make libwhisper.a -j

Then install the requirements:

• openblas library
• ffmpeg

Command to install on certain distros:

# Debian / Ubuntu
sudo apt-get install libopenblas-dev ffmpeg

# Arch Linux
pacman -S openblas ffmpeg

# Search it yourself for other distros

Then install manually the smart-whisper module:

BYOL="$PWD/smart-whisper/whisper.cpp/libwhisper.a -lopenblas" npm install smart-whisper

And install the rest of the modules:

npm install