MP3net: coherent, minute-long music generation from raw audio with a simple convolutional GAN

TL;DR

MP3net introduces a convolutional GAN that generates long, coherent, high-quality music from raw audio by leveraging MP3 compression techniques and psychoacoustic principles, enabling near-instantaneous song creation.

Contribution

The paper presents a novel deep convolutional GAN architecture that incorporates MP3 compression insights and psychoacoustic modeling for coherent, long-duration music generation from raw audio.

Findings

Generated 95-second stereo tracks at 22kHz after 250 hours of training.

Model achieves long-range coherence in music samples.

Generation process is nearly instantaneous.

Abstract

We present a deep convolutional GAN which leverages techniques from MP3/Vorbis audio compression to produce long, high-quality audio samples with long-range coherence. The model uses a Modified Discrete Cosine Transform (MDCT) data representation, which includes all phase information. Phase generation is hence integral part of the model. We leverage the auditory masking and psychoacoustic perception limit of the human ear to widen the true distribution and stabilize the training process. The model architecture is a deep 2D convolutional network, where each subsequent generator model block increases the resolution along the time axis and adds a higher octave along the frequency axis. The deeper layers are connected with all parts of the output and have the context of the full track. This enables generation of samples which exhibit long-range coherence. We use MP3net to create 95s stereo tracks with a 22kHz sample rate after training for 250h on a single Cloud TPUv2. An additional benefit of the CNN-based model architecture is that generation of new songs is almost instantaneous.