A bass tab generator using a LSTM approach
The purpose of this Artificial Intelligence project is to create bass riffs based on playing patterns from different bands and musics from the funk rock genre.
This software depends on the following third party libraries to be executed:
- TensorFlow: version 2.0 or later
$ pip install tensorflow- PyGuitarPro: version 0.6 or later
$ pip install PyGuitarProFirst of all, I would like to point out that when referring to the .gpx format, I am actually referring to the Guitar Pro .gp3, .gp4 and .gp5 formats
The execution of the scripts is done exclusively in the terminal. Below are their descriptions.
This script makes it possible to rip the bass track from .gpx files in a certain directory.
python bass_ripper.py [input dir] [output dir]ALWAYS perform the rip of the bass tracks before converting any file.
This script, as its name suggests, will be responsible for converting files. There are two execution methods, from .gpx to .txt and from .txt to .gpx.
Conversion to .txt:
Specifying an output file is optional, the default output is "output.txt".
python converter.py [-t or --txt] [input dir] [output file]The generated file will be written as follows: (see PyGuitarPro docs for better understanding)
- Beat
b D OT NMA line that represents a beat will always start with a b, and will have the five following parameters:
- D: duration of the beat
- O: specifies if the note is dotted
- T: type of the beat
- E: tuplet enters
- M: tuplet tempo
- Note
n S N THLA line that represents a note will always start with a n, and will have the five following parameters:
- S: bass string
- N: played note
- T: note type
- H: presence of hammer-on
- L: presence of slide
Conversion to .gpx:
Specifying an output file is optional, the default output is "output.gp5". Note that it is preferable to generate files with the .gp5 extension
python converter.py [-g or --gpx] [input file] [output file]