StudyLens

Turn Any Video Into Complete Study Notes — Instantly

Paste a YouTube link or upload a video. Get transcripts, summaries, quizzes, Q&A, and more.
Transcription runs 100% locally. Only AI generation needs an API key.

Getting Started · Features · How It Works · API Keys

---

The Problem

You watch a 2-hour lecture. You take notes. You miss things. You rewatch parts. You spend more time organizing notes than actually studying.

StudyLens fixes this. Paste a link, pick what you want, and get study-ready material in minutes — not hours.

---

What You Get

From any video, generate: Output What it does Transcript Full text with clickable timestamps Summary Notes Structured, concise study notes Main Topics Topic-by-topic breakdown with key points Detailed Q&A Exam-style questions with model answers Practice Questions Short answer, long answer, and true/false MCQ Quiz 15-20 multiple choice questions Exhaustive Notes Every single detail captured

Plus: Feature Description Chat with Notes Highlight text, ask AI to explain it Generate More Add new output types later without re-processing Merge Sessions Combine multiple videos into one study guide Fullscreen Reader Clean reading view with zoom (60%-200%) Dark Mode Toggle between light and dark themes User Accounts Save history, chats, and settings

---

How It Works

  YouTube URL / Video File
          |
          v
  yt-dlp + ffmpeg              Extracts audio locally
          |
          v
  faster-whisper               Transcribes on YOUR machine (no cloud)
          |
          v
  LLM API                      Generates study materials
  (OpenAI / Gemini / Claude)   (only step that needs an API key)
          |
          v
  SQLite                       Saves everything for logged-in users

No subscriptions. No monthly fees. Transcription is fully local using faster-whisper. The only cost is the LLM API call for generating summaries/quizzes — typically a few cents per video using GPT-4o-mini.

---

Tech Stack

Component	Technology	Runs Locally?
Transcription	faster-whisper / mlx-whisper (auto-detected)	Yes
Video Download	yt-dlp + ffmpeg	Yes
AI Generation	OpenAI gpt-4o-mini / Google gemini-2.5-flash / Anthropic claude-sonnet-4-6	API call
Backend	Python, FastAPI, Uvicorn	Yes
Frontend	Vanilla JavaScript, Tailwind CSS, Marked.js	Yes
Database	SQLite (WAL mode)	Yes
Auth	PBKDF2-HMAC-SHA256, secure httponly cookies	Yes

Everything runs on your machine except the LLM API calls for generating AI content.

Transcription Speed

Transcription speed depends on your hardware. StudyLens auto-detects and uses the fastest option available:

Hardware	Backend Used	1-hour video
Apple Silicon Mac (M1/M2/M3/M4)	mlx-whisper (Metal GPU)	~5-8 min
NVIDIA GPU (CUDA)	faster-whisper (GPU)	~3-5 min
CPU only (Intel Mac, Docker)	faster-whisper (CPU)	~15-20 min

No configuration needed — the right backend is selected automatically. On Mac, run.sh installs mlx-whisper for you.

---

Getting Started

Run with Docker — One Command, No Setup

Install Docker Desktop (free), then pick your AI provider and run one command:

Provider	Command
OpenAI gpt-4o-mini Get key	docker run -d -p 8000:8000 \ -e OPENAI_API_KEY=sk-your-key-here \ -v studylens-data:/app/data \ --restart unless-stopped \ --name studylens \ tusharkhatriofficial/studylens
Google Gemini gemini-2.5-flash Get key	docker run -d -p 8000:8000 \ -e GEMINI_API_KEY=your-gemini-key-here \ -v studylens-data:/app/data \ --restart unless-stopped \ --name studylens \ tusharkhatriofficial/studylens
Anthropic claude-sonnet Get key	docker run -d -p 8000:8000 \ -e ANTHROPIC_API_KEY=sk-ant-your-key-here \ -v studylens-data:/app/data \ --restart unless-stopped \ --name studylens \ tusharkhatriofficial/studylens

Open http://localhost:8000 and start studying.

You only need one key. Pick whichever provider you prefer. Google Gemini has a free tier.

Your data is safe

The -v studylens-data:/app/data part stores your SQLite database (accounts, study history, chats) in a persistent Docker volume. Your data survives:

• Stopping the container (docker stop studylens)
• Restarting Docker Desktop
• Updating to a new version
• Even deleting and re-creating the container

The only way to lose data is explicitly deleting the volume with docker volume rm studylens-data.

Common Docker commands

What	Command
Stop	docker stop studylens
Start again	docker start studylens
View logs	docker logs studylens
Update to latest	docker pull tusharkhatriofficial/studylens && docker rm -f studylens then run the original command again
Remove everything	docker rm -f studylens && docker volume rm studylens-data

---

Docker Compose setup (alternative)

Create a docker-compose.yml file:

services:
  studylens:
    image: tusharkhatriofficial/studylens:latest
    ports:
      - "8000:8000"
    environment:
      - OPENAI_API_KEY=sk-your-key-here    # or GEMINI_API_KEY or ANTHROPIC_API_KEY
    volumes:
      - studylens-data:/app/data
    restart: unless-stopped

volumes:
  studylens-data:

Then: docker compose up -d

Manual setup without Docker (for developers)

Requirement	Install
Python 3.10+	python.org
ffmpeg	brew install ffmpeg (macOS) / apt install ffmpeg (Linux / WSL)
An LLM API key	See API Key Setup below

git clone https://github.com/tusharkhatriofficial/study_lense.git
cd study_lense
cp .env.example .env     # then edit .env and add your key
chmod +x run.sh
./run.sh

Open http://localhost:8000.

First run creates a virtual environment, installs dependencies, and downloads the Whisper base model (~150MB). One-time setup.

---

API Key Setup

StudyLens needs an LLM API key only for generating AI content (summaries, quizzes, Q&A, etc.). Transcription is fully local and free.

Where to get a key

Provider	Model Used	Get Key
OpenAI	gpt-4o-mini	platform.openai.com/api-keys
Google Gemini	gemini-2.5-flash	aistudio.google.com/apikey
Anthropic	claude-sonnet-4-6	console.anthropic.com/settings/keys

You only need one key. Pick whichever provider you prefer.

How keys work in StudyLens

Setup	What happens
Key in .env file	Acts as the server default. All users can use it (with rate limits).
Key in Settings page	Users add their own key in-app. Overrides the server default. No rate limits.
No key at all	Transcription still works. AI features ask the user to add a key in Settings.

Rate limits (when using the server default key)

User Type	Video Processing	AI Chat
Guest (no account)	3/day	5/day
Logged-in user	20/day	30/day
Using own API key	Unlimited	Unlimited

---

Features in Detail

Generate More — add output types later without re-processing

Generated only a transcript and summary? Open that session from history, and you'll see checkboxes for every output type you haven't generated yet (MCQ, Q&A, Topics, etc.). Check what you want and hit Go. StudyLens uses the saved transcript — no need to re-download or re-transcribe the video.

Chat with Notes — highlight text and ask AI about it

In any study session, highlight text in your notes. A tooltip appears letting you ask AI about the selection. The AI has full context of your transcript and all generated outputs, so answers are specific to your content.

There's also a standalone full-screen chat that can optionally reference any study session.

Merge Sessions — combine multiple videos into one guide

Select 2 or more study sessions from your history and merge them. StudyLens combines all transcripts (labeled by source) and generates new study materials from the combined content. Great for combining lecture series or related videos.

Multi-Provider AI — choose your preferred LLM

StudyLens auto-detects the provider from your API key format:

• Keys starting with sk- → OpenAI (gpt-4o-mini)
• Keys starting with AIza → Google Gemini (gemini-2.5-flash)
• Keys starting with sk-ant- → Anthropic Claude (claude-sonnet-4-6)

You can set a different key per provider in Settings and switch between them.

---

Project Structure

study_lense/
├── backend/
│   ├── main.py           # FastAPI — all API endpoints, SSE progress, auth
│   ├── db.py             # SQLite — users, history, chats, usage tracking
│   ├── downloader.py     # yt-dlp + ffmpeg — download & extract audio
│   ├── transcriber.py    # faster-whisper — local transcription
│   └── summarizer.py     # Multi-provider LLM — content generation
├── static/
│   ├── index.html        # Single-page app
│   ├── app.js            # Frontend logic (vanilla JS)
│   └── style.css         # Tailwind CSS + custom styles
├── data/                  # SQLite database (gitignored)
├── .env                   # Your API keys (gitignored)
├── Dockerfile
├── requirements.txt
└── run.sh

---

API Reference

Click to expand

Video Processing

Method	Endpoint	Description
POST	/api/process	Process a YouTube URL or uploaded video file
GET	/api/status/{task_id}	SSE stream — real-time progress updates

Study Sessions

Method	Endpoint	Description
GET	/api/history	List all saved study sessions
GET	/api/history/{id}	Get full session (transcript, outputs, metadata)
POST	/api/history/{id}/generate-more	Generate additional output types
PATCH	/api/history/{id}	Rename a session
DELETE	/api/history/{id}	Delete a session
POST	/api/merge	Merge multiple sessions into one

Chat

Method	Endpoint	Description
GET	/api/history/{id}/chats	List chats for a study session
POST	/api/history/{id}/chats	Start a new chat (with optional text selection)
POST	/api/chats/{id}/message	Send message to existing chat
POST	/api/standalone-chat	Full-screen chat (optionally references a session)

Auth & Settings

Method	Endpoint	Description
POST	/api/register	Create account
POST	/api/login	Login
POST	/api/logout	Logout
GET	/api/me	Get current user
POST	/api/keys	Save an API key
GET	/api/health	Health check (also reports if a default key is configured)

---

Contributing

Contributions welcome. Open an issue or submit a PR.

License

MIT

---

Built by Tushar Khatri