unosd / telegramstt

Create a Telegram bot to transcribe audio messages

TelegramSTT

Creates a Telegram bot to transcribe audio messages.

It uses an Azure Logic App invoked by the Telegram BOT webhook that in turn calls the Azure Speech services to transcribe the audio and the bot responds with the transcribed message.

How to create a Telegram bot

From Telegram, contact @BotFather

/newbot <name your bot> <write your unique Telegram handle for the bot>

Now copy the bot auth key returned.

pulumi config set azure-native:location "West Europe"
pulumi config set bot-handle "<your bot handle from above>"
pulumi config set bot-key "<your bot key returned by @BotFather>"
pulumi up
# Currently this line won't work as the trigger API invoke args do not take the trigger name, getting URL from the LA in the portal
curl -X POST -H "Content-Type: application/json" -d "{\"url\":\"$(pulumi stack output WorkflowTriggerUrl --show-secrets)\", \"allowed_updates\":[\"message\"]}" "https://api.telegram.org/bot$(pulumi config get bot-key)/setWebhook"

How to use it

Forward any audio message in a conversation to your bot and it will respond with the transcription.