Catch-A-Waveform: Learning to Generate Audio from a Single Short Example

Abstract

Models for audio generation are typically trained on hours of recordings. Here, we illustrate that capturing the essence of an audio source is typically possible from as little as a few tens of seconds from a single training signal. Specifically, we present a GAN-based generative model that can be trained on one short audio signal from any domain (e.g. speech, music, etc.) and does not require pre-training or any other form of external supervision. Once trained, our model can generate random samples of arbitrary duration that maintain semantic similarity to the training waveform, yet exhibit new compositions of its audio primitives. This enables a long line of interesting applications, including generating new jazz improvisations or new a-cappella rap variants based on a single short example, producing coherent modifications to famous songs (e.g. adding a new verse to a Beatles song based solely on the original recording), filling-in of missing parts (inpainting), extending the bandwidth of a speech signal (super-resolution), and enhancing old recordings without access to any clean training example. We show that in all cases, no more than 20 seconds of training audio commonly suffice for our model to achieve state-of-the-art results. This is despite its complete lack of prior knowledge about the nature of audio signals in general.Preprint. Under review.

Draw and Listen! A Sketch-Based System for Music Inpainting

“…In the audio domain, the inpainting methods try to recover missing data in the waveform, which can occur due to various reasons such as distortions and transmission errors (Adler et al, 2012;Marafioti et al, 2020). The same term has also been used to describe the bandwidth extension problem, where the missing high frequency content has to be estimated (inpainted) from the low frequencies (Greshler et al, 2021).…”

Taming Visually Guided Sound Generation

“…Previous works have proposed models to controllably generate e.g. images [13,17,38,45,48,51,55,57,73,76,77], videos [6,12,25,37,42,46,64,65,65,71], and audios [1,9,15,22,24,47,62,63], or separate sounds [18,19,79,80,84]. However, most of the audio works are music-related, and only a few attempts have been made to generate visually guided audio in an open domain setup [11,83].…”