Telegram bot based on Python 3 for running Speech-To-Text (STT) and Text-To-Speech (TTS) queries. Languages supported: Russian, English (queries and interface).

Module for working with Telegram API: Aiogram.

Software for converting audio files into different formats: FFmpeg.

STT and TTS queries are performed using the following libraries:

• Vosk (STT)

• Pyttsx3 (TTS)

Bot supports two voices (male and female), whose names are set in the configuration file.

Bot has a special function called extra_text_processing, which introduces additional methods of processing text received from Vosk. By going through it, the text can be made more human and correct in terms of writing. The materials required for this function are stored strictly in the src/etp.

Note: Created and tested on Windows platform, Python 3.11.4

The following steps are required for RantoVox to work correctly:

1. Clone the repository (download source code)

2. Create a virtual environment using python -m venv venv and activate it

3. Install dependencies using pip with requirements.txt

4. Download latest vosk russian and english language models (the small model is more preferable), drop them into src/lang (You can use make download to automatically download and post, requires curl and tar)

5. Create your own .env file in root folder with variables described in Environment file section.

git clone https://github.com/Ggorets0dev/rantovox-telegram-bot.git
cd rantovox-telegram-bot
pip install -r requirements.txt

The following commands are available in RantoVox:

• start - Launch a bot for your account

• help - Get an informational summary of the operating principles

• setlocale - Set language of bot's interface

• setvoice - Set voice gender for requests (TTS)

• setlang - Set language for requests (STT)

A .env file with the following variables must be created before running the bot:

Note: Default - value that the bot will take on its own if the value is in the wrong format in the environment file

Note: Call the bot for a list of available votes, filling in all remaining variables. It will display a list of available values (be careful: not all voices support Russian and English at the same time)

Bot deletes all temporary files immediately after a TTS or STT request. All conversion is done on the host with the help of the libraries described above. Only user's Login and ID are recorded in logs when requesting, composition of request remains hidden to host.