Neural Midi Synthesizer

The code for the research work on using simplified DDSP approach for style-aware modeling of the sound of real instruments.

Find more details, including links to the abstract and slides, as well as the recording of my talk at DMRN+16, here: "Building style-aware neural MIDI synthesizers using simplified differentiable DSP approach".

This repository was originally a clone of https://github.com/raraz15/ddsp_simplified. Refer to the readme of the original repo to see some comments on the baseline implementation.

Name		Name	Last commit message	Last commit date
Latest commit History 83 Commits
Notebooks		Notebooks
audio_clips/Violin		audio_clips/Violin
cache		cache
configs		configs
dsp_utils		dsp_utils
tests		tests
utils		utils
.gitignore		.gitignore
README.md		README.md
callbacks.py		callbacks.py
config_key_constants.py		config_key_constants.py
dataloader.py		dataloader.py
decoders.py		decoders.py
encoders.py		encoders.py
feature_extraction.py		feature_extraction.py
feature_names.py		feature_names.py
losses.py		losses.py
metrics.py		metrics.py
misc_constants.py		misc_constants.py
models.py		models.py
postprocessing.py		postprocessing.py
preprocessing.py		preprocessing.py
pretrained_crepe.py		pretrained_crepe.py
requirements_cpu.txt		requirements_cpu.txt
resnet.py		resnet.py
synthesis_by_midi_lib.py		synthesis_by_midi_lib.py
synthesize_from_midi.py		synthesize_from_midi.py
synthesize_from_midi_lib.py		synthesize_from_midi_lib.py
synthesizers.py		synthesizers.py
timbre_transfer.py		timbre_transfer.py
train_supervised.py		train_supervised.py
train_unsupervised.py		train_unsupervised.py
train_utils.py		train_utils.py
utilities.py		utilities.py

Provide feedback