MelGAN-VC: Voice Conversion and Audio Style Transfer on arbitrarily long samples using Spectrograms
This addresses the challenge of voice conversion in real-world scenarios where parallel data is scarce, though it is incremental as it builds on existing GAN-based methods.
The authors tackled the problem of voice conversion without requiring parallel recordings by proposing MelGAN-VC, which uses spectrograms and a GAN with a siamese network to convert audio of arbitrary length, achieving successful results on clean and noisy speech as well as music style transfer.
Traditional voice conversion methods rely on parallel recordings of multiple speakers pronouncing the same sentences. For real-world applications however, parallel data is rarely available. We propose MelGAN-VC, a voice conversion method that relies on non-parallel speech data and is able to convert audio signals of arbitrary length from a source voice to a target voice. We firstly compute spectrograms from waveform data and then perform a domain translation using a Generative Adversarial Network (GAN) architecture. An additional siamese network helps preserving speech information in the translation process, without sacrificing the ability to flexibly model the style of the target speaker. We test our framework with a dataset of clean speech recordings, as well as with a collection of noisy real-world speech examples. Finally, we apply the same method to perform music style transfer, translating arbitrarily long music samples from one genre to another, and showing that our framework is flexible and can be used for audio manipulation applications different from voice conversion.