AudioFP - Audio Intelligence Platform

Google-like audio search for your private audio library.
Upload a 3-second clip and instantly find where it appears across thousands of files.

Audio Finger Print is a production-ready, self-hosted audio fingerprinting platform. It combines a high-accuracy Shazam-style matching engine with a clean web interface, letting you build a searchable audio knowledge base from your own files - no cloud, no transcription, no API keys required.

---

Features

• Short-clip search - recognise audio from clips as short as 3 seconds, even with background noise
• Exact & near-match - Shazam-style spectral fingerprinting with time-offset voting
• Match timestamp - tells you exactly where in the original recording the clip comes from
• Batch indexing - upload individual files or point to a local folder; indexing runs in the background
• Fast search - batch hash lookups in a single SQL query; SQLite with WAL mode, covering index, and 64 MB page cache
• 500 MB upload limit - handles large uncompressed WAV files out of the box
• Simple UI - drag-and-drop search, library management, and live stats in one page
• REST API - every feature is accessible via a JSON API for programmatic use

---

UI Screenshots

Quick Start

Prerequisites - install FFmpeg (or SoX) so the platform can decode all supported audio formats:

# macOS
brew install ffmpeg

# Ubuntu / Debian
sudo apt-get install ffmpeg

# Windows (via Chocolatey)
choco install ffmpeg

# Windows (via Scoop)
scoop install ffmpeg

Verify the installation:

ffmpeg -version

Then:

# 1. Clone and install dependencies
git clone https://github.com/your-org/audio-fingerprint.git
cd audio-fingerprint
pip install -r requirements.txt

# 2. Start the server
python run.py

# 3. Open the browser
# → http://localhost:5000

The server creates data/fingerprint_dev.db on first run. Your library persists across restarts.

Options

python run.py --port 8080                   # custom port
python run.py --env production --port 80    # production config

---

REST API

The full API lives under /api/v1. See docs/API.md for complete reference.

---

Configuration

Variable	Default	Description
STORAGE_TYPE	sqlite	memory · sqlite · postgres
SQLITE_DATABASE_PATH	data/fingerprint_dev.db	SQLite file path
UPLOAD_FOLDER	data/uploads	Where uploaded files are saved
SAMPLE_RATE	11025	Audio sample rate (Hz)
FAN_VALUE	10	Hash pairs per anchor peak (higher = more robust)
MAX_CONTENT_LENGTH	500 MB	Maximum upload size

Edit config/development.py (dev) or config/production.py (prod) to change defaults.

---

Architecture

The platform has two main data flows:

1. Training & Indexing - Audio files are loaded, preprocessed, passed through spectral peak extraction and hash generation, then stored in the fingerprint database.
2. Application & Search - User query clips follow the same fingerprinting pipeline, and the matcher queries the database to score and return the best matches via the Flask REST API.

See docs/ARCHITECTURE.md for the full algorithm walkthrough.

---

License

MIT - see LICENSE.