We want to create some tools that wrap around audio recognizing with dejavu.

To make it easy to add songs to our database, we have spent time making a program, that you give a search query (song - artist), which it then searches for, downloads, converts to mp3 and fingerprints into the database. Then it can be recognized afterwards.

We will make the changes to the dejavu project that makes it more efficient - Change timed searches to confidence level based searches

We recommend running the project on a local MySQL Database. Create a database, and a user with permissions for that Database, and put in your settings in dejavu.cnf

Currently only supported on Linux/Mac:

git clone https://github.com/kasperonfire/ImpossibleTechnologyProject
cd ImpossibleTechnologyProject
chmod +x DejavuSetup.sh
./DejavuSetup.sh

Full List of all dependencies, if something should fail: apt:

python3
mysql-server 
python3-mysql.connector 
python3-pip 
python3-tk
portaudio19-dev
ffmpeg

Python libraries:

numpy
pydub
wavio
matplotlib
scipy
pyaudio
pyqtgraph
youtube-dl

To Download a song to learn directory:

python dejavu/youtube.py -v "search query - can be one word, or artist + title, or anything"

To Learn the songs in the learn directory:

python dejavu/learner.py

To Recognize a song:

python dejavu/visualize_pyqtgraph.py

Dejavu audio fingerprinting

Article about dejavu

youtube-dl