VoiceBot

Telegram bot that transcribes voice messages, video notes, audio files, and music into text using Groq's Whisper API.

Try VoiceBot on Telegram!

Quick Start

Send or forward any voice message, video note, audio track, or document to @egvoicebot to receive a transcription.

---

Features

• Supports voice messages alongside any video file format.
• Uses Groq whisper-large-v3 for speech-to-text processing.
• Interactive inline show/hide buttons for output text.
• Supports files up to 2GB via local Telegram Bot API server integration.

---

How to Run Locally

Requirements

• Groq API Key
• Telegram Bot Token

Setup

1. Clone the repository:

   git clone https://github.com/egraich/VoiceBot.git
   cd VoiceBot

2. Create a .env file inside VoiceBot/:

   BOT_TOKEN=your_telegram_bot_token
   GROQ_API_KEY=gsk_your_groq_api_key_here
   ADMIN_ID=your_admin_id
   TEMP_DIR=path/to/temp/files/directory
   BOT_API_URL=https://api.telegram.org

3. Install requirements:

   pip install -r requirements.txt

4. Run VoiceBot:

   python main.py

---

How It Works

1. Media Ingestion & Audio Extraction: Incoming files are downloaded locally. Video files are converted to 16kHz mono FLAC via FFmpeg before sending to the API.
2. Transcription: Audio files are transmitted asynchronously to the Groq Cloud API using the whisper-large-v3 model.
3. Storage & Local API: Connecting the bot to a local Telegram Bot API server removes the default 20MB file size ceiling, allowing uploads up to 2000MB. Caching and stats are stored in SQLite database file via aiosqlite using PRAGMA journal_mode=WAL to handle concurrent operations without database locks.

---

Credits

• Aiogram 3 — Async Telegram Bot API framework.
• Groq API — Fast speech recognition endpoint.
• FFmpeg — Multimedia processing framework.
• aiosqlite — Async SQLite libriary for Python.

Made by egraich