freespeech

Installation

Prerequisites:

• Ffmpeg
• (Optional) Docker

Clone the Repo

gh repo clone astaff/freespeech
cd freespeech

or

git clone https://github.com/astaff/freespeech.git
cd freespeech

Setup the Credentials

1. Copy .env file into /
2. mkdir id and copy credentials (Google, etc) to id/

Setup Git hooks

Run bin/copy_hooks.sh

Common tasks

From project home directory:

• To run locally: pip install -e . and freespeech --help
• To test locally: pip install -e ".[test]"
• To run the tests: make test
• To generate training data: make data
• To build docs locally: pip install -e ".[docs]" && make docs

Cloud Info

Azure

Currently, text-to-speech from Azure is used. In order to get a working environment, one needs to:

1. Have a Microsoft Azure account. https://azure.microsoft.com/en-us/free/ to get started
2. Create a Speech resource in certain Azure region https://ms.portal.azure.com/#create/Microsoft.CognitiveServicesSpeechServices
3. Obtain values for AZURE_REGION and AZURE_SUBSCRIPTION_KEY required by the app link to guide

General quickstart guide: https://docs.microsoft.com/en-us/azure/cognitive-services/speech-service/get-started-speech-to-text

Google

TBD

Telegram

Service is available as a chatbot. You would need to create a chat bot via https://t.me/BotFather (and hence obtain TELEGRAM_BOT_TOKEN to set env). Please also disable privacy mode for bot since we want to react to plain mentions in groups. This can be done in BotFather settings for your bot via telegram itself.

TELEGRAM_WEBHOOK_URL should also be set. This should be a publicly accessible URL pointing to the webhook. Telegram would push messages to this URL. Typically this would look like http://address.tld:8080/tg_webhook when ran with start-telegram on address.tld host.

You can use ngrok to forward local port from development machine to a public address.