ASR/CV Telegram Bot

Usage

Prepare the environment. Note that parameters in
grpc_audio_client.py are adjusted to quartznet15x5 model.

./fetch_models.sh  # CV models; ASR ─ you load manually (onnx)
pip install requests aiogram  # telegram bot API dependencies

Create a Telegram bot, obtain its token (all can be done via BotFather).
Use ./start.sh to launch triton-server and fastapi "dummy client".
You can also check the script options with ./start.sh -h command.
Pass the token to the script launching the telegram bot
```
python voice_transcription_bot.py -t YOUR_API_TOKEN
```

triton-inference-server ─ some files from there (grpc_image_client.py and fetch_models.sh)
are adjusted to the use in this repository.

Name		Name	Last commit message	Last commit date
Latest commit History 3 Commits
docker		docker
LICENSE		LICENSE
README.md		README.md
asr_utils.py		asr_utils.py
dummy_client_api.py		dummy_client_api.py
fetch_models.sh		fetch_models.sh
grpc_audio_client.py		grpc_audio_client.py
grpc_image_client.py		grpc_image_client.py
start.sh		start.sh
voice_transcription_bot.py		voice_transcription_bot.py