Telegram bot that transcribes audio messages sent to it.
The bot keeps no state and deletes files after processing.
PSA: this is under heavy development
git clone https://github.com/bolt-observer/transcriber-bot.git
cd transcriber-bot
pip install -r requirements.txt
cp example_settings.py settings.py
# edit settings.py
python transcriberbot.py
TELEGRAM_BOT_TOKEN
- Follow telegram's instruction on how to create your own telegram bot. This will provide you with a token that you put in settings.py
MODEL_PATH
- Download one of the whisper models converted to ggml format from HuggingFace and put the path to the file in settings.py
. Example MODEL_PATH="~/whisper.cpp/models/ggml-base.en.bin"
- audio file filtering
- inline commands so it can transcribe voice msgs in groups
- better error handling