🎙️ TikTok & Instagram Transcription Bot

A Telegram bot that downloads TikTok and Instagram videos and transcribes their audio — 100% free, runs locally using OpenAI Whisper.

How it works

1. User sends a TikTok or Instagram link to the bot
2. Bot downloads the audio using yt-dlp (free, open source)
3. Bot transcribes it using OpenAI Whisper (runs on your machine, no API key needed)
4. Bot replies with the transcription text

---

🛠️ Setup

Prerequisites

• Python 3.9+
• ffmpeg installed on your system
• A Telegram Bot Token (free, from @BotFather)

Install ffmpeg

Ubuntu/Debian:

sudo apt install ffmpeg

macOS:

brew install ffmpeg

Windows: Download from https://ffmpeg.org/download.html and add to PATH.

---

Step 1: Create your Telegram bot

1. Open Telegram and search for @BotFather
2. Send /newbot
3. Follow the prompts to name your bot
4. Copy the API token you receive

---

Step 2: Install dependencies

# Clone or download this project
cd tiktok_ig_transcriber_bot

# Create a virtual environment (recommended)
python -m venv venv
source venv/bin/activate   # Windows: venv\Scripts\activate

# Install Python packages
pip install -r requirements.txt

⚠️ First install will download PyTorch (~1GB). This is a one-time download.

---

Step 3: Set your bot token

# Copy the example env file
cp .env.example .env

# Edit .env and add your bot token
# TELEGRAM_BOT_TOKEN=123456789:ABCdef...

Then load it before running:

# Linux/Mac
export TELEGRAM_BOT_TOKEN="your_token_here"

# Windows (Command Prompt)
set TELEGRAM_BOT_TOKEN=8613704159:AAGxAKLobIzv6lRawRkilGc-hyUoVmrBG0M

# Windows (PowerShell)
$env:TELEGRAM_BOT_TOKEN="your_token_here"

---

Step 4: Run the bot

python bot.py

You should see:

INFO - Whisper model loaded.
INFO - Bot is running...

Now open Telegram, find your bot, and send it a TikTok or Instagram link!

---

🐳 Docker (optional, easier deployment)

# Build the image (includes ffmpeg + pre-downloaded Whisper model)
docker build -t transcriber-bot .

# Run it
docker run -e TELEGRAM_BOT_TOKEN="your_token_here" transcriber-bot

---

⚙️ Configuration

Environment Variable	Default	Description
TELEGRAM_BOT_TOKEN	(required)	Your bot token from BotFather
WHISPER_MODEL	base	Whisper model size (see below)

Whisper model sizes

Model	RAM needed	Speed	Accuracy
tiny	~1 GB	Fastest	Basic
base	~1 GB	Fast	Good ✅
small	~2 GB	Medium	Better
medium	~5 GB	Slow	Great
large	~10 GB	Slowest	Best

For most use cases, base is the sweet spot.

---

🚀 Free hosting options

Want to run this 24/7 for free?

• Render.com — Free tier, deploy with Docker
• Railway.app — Free tier available
• Fly.io — Free tier with enough resources for tiny/base model
• Oracle Cloud Free Tier — Permanent free VMs (ARM, 4 cores, 24GB RAM)
• Your own PC/Raspberry Pi — Run it at home

⚠️ Note: Cloud free tiers often have limited RAM. Use WHISPER_MODEL=tiny for tighter limits.

---

⚠️ Limitations

• Private videos cannot be downloaded (must be public)
• Instagram sometimes blocks downloads depending on the account's privacy settings
• Very long videos (10+ min) may take a few minutes to transcribe
• TikTok occasionally changes its download mechanism; update yt-dlp if issues arise:

  pip install -U yt-dlp

---

📦 Dependencies

Package	Purpose	Cost
python-telegram-bot	Telegram bot framework	Free
yt-dlp	Download TikTok/IG video/audio	Free
openai-whisper	Speech-to-text transcription	Free (runs locally)
ffmpeg	Audio processing	Free