> ## Documentation Index
> Fetch the complete documentation index at: https://docs.nometria.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Screen Recorder Analyzer

> Analyze screen recordings with Whisper audio transcription, Tesseract OCR on keyframes, and GPT action extraction.

`screen-recorder-analyzer` extracts a chronological list of what a user was doing from any screen recording. Combines Whisper audio transcription, Tesseract OCR on keyframes, and GPT action extraction.

<CardGroup cols={2}>
  <Card title="GitHub" icon="github" href="https://github.com/nometria/screen-recorder-analyzer">
    nometria/screen-recorder-analyzer
  </Card>

  <Card title="PyPI" icon="python" href="https://pypi.org/project/screen-recorder-analyzer/">
    screen-recorder-analyzer on PyPI
  </Card>
</CardGroup>

## Install

```bash theme={null}
# System dependencies (required)
brew install ffmpeg tesseract          # macOS
sudo apt install ffmpeg tesseract-ocr  # Ubuntu

# Install base package
pip install screen-recorder-analyzer

# Install with all OCR and audio engines
pip install screen-recorder-analyzer[full]
```

## Usage

```bash theme={null}
# Set API key
export OPENAI_API_KEY=sk-proj-...

# Analyze a recording
screen-analyze demo.mp4

# JSON output (for piping or programmatic use)
screen-analyze demo.mp4 --format json

# Run as REST API
screen-analyze --api --port 8000
# POST /analyze  { "file": "path/to/recording.mp4" }
```

## Output format

```json theme={null}
[
  {
    "id": "1",
    "timestamp": "0:00",
    "tools": ["excel"],
    "action": ["viewing spreadsheet data"]
  },
  {
    "id": "2",
    "timestamp": "0:15",
    "tools": ["hubspot"],
    "action": ["navigating CRM", "viewing contacts list"]
  },
  {
    "id": "3",
    "timestamp": "0:42",
    "tools": ["gmail"],
    "action": ["composing email", "sending email"]
  }
]
```

## Pipeline

| Step                  | Engine                    | What it does                                 |
| --------------------- | ------------------------- | -------------------------------------------- |
| **Frame extraction**  | ffmpeg                    | Extracts keyframes at configurable intervals |
| **OCR**               | Tesseract → EasyOCR → GCV | Reads on-screen text from each frame         |
| **Transcription**     | Whisper                   | Transcribes audio narration                  |
| **Action extraction** | GPT-4                     | Combines OCR + audio → structured actions    |

## Optional engines

```bash theme={null}
# EasyOCR + OpenCV (GPU-accelerated secondary OCR)
pip install screen-recorder-analyzer[gpu]

# Google Cloud Vision (fallback OCR engine)
pip install screen-recorder-analyzer[gcp]
```

## Use cases

* **User research**: Understand what testers are doing in session recordings
* **QA automation**: Extract action sequences from recorded test runs
* **Onboarding analysis**: See where users get stuck in screen recordings
* **Compliance logging**: Generate structured audit trails from recorded sessions