telegram-azure-transcribe-bot

Simple Telegram bot to transcribe voice messages sent to it via Azure speech-to-text.

Configure

You'll need a free (or paid) Azure account with a Speech Services resource. You can visit the Azure Speech Services page to get started. Make sure to select Free Plan when creating the resource and NOT the pay-as-you-go option.

Configure .env file with the following:

• TELEGRAM_BOT_TOKEN: Token provided by @BotFather
• TELEGRAM_BOT_ALLOWED_CHAT_IDS: optional comma-separated list of chat IDs that may use this bot
• SPEECH_KEY: speech key from Azure speech service
• SPEECH_REGION: region from Azure speech servce
• LOG_LEVEL: optional, use strings from Python 'logging' such as INFO, WARN, DEBUG, ERROR

Run in Docker

You have two options for running in Docker.

Use pre-built package

docker pull ghcr.io/ceralor/telegram-azure-transcribe-bot:latest

Build locally

Clone the repository, then use `docker build . -t

Running

Either create a .env file with the required variables (below) or supply via -e LOG_LEVEL=WARN in the command line.

docker run --restart unless-stopped --env-file .env --detach ghcr.io/ceralor/telegram-azure-transcribe-bot:latest

Running bare or building

Requires libraries:

• python-telegram-bot[job-queue]
• python-dotenv
• azure-cognitiveservices-speech
• pydub

Requires installed:

• dotnet
• ffmpeg

Then python bot.py