Melgan: Generative Adversarial Networks For Conditional Waveform Synthesis

Abstract

Previous works (Donahue et al., 2018a; Engel et al., 2019a) have found that generating coherent raw audio waveforms with GANs is challenging. In this paper, we show that it is possible to train GANs reliably to generate high quality coherent waveforms by introducing a set of architectural changes and simple training techniques. Subjective evaluation metric (Mean Opinion Score, or MOS) shows the effectiveness of the proposed approach for high quality mel-spectrogram inversion. To establish the generality of the proposed techniques, we show qualitative results of our model in speech synthesis, music domain translation and unconditional music synthesis. We evaluate the various components of the model through ablation studies and suggest a set of guidelines to design general purpose discriminators and generators for conditional sequence synthesis tasks. Our model is non-autoregressive, fully convolutional, with significantly fewer parameters than competing models and generalizes to unseen

Stats

• citations0
• S2 citations—
• github stars0
• HF likes0
• heat score0.00
• arxiv keykumar2019melgan

Related papers

• Chunked Autoregressive GAN For Conditional Waveform Synthesis (2021)0.00
• Adversarial Audio Synthesis (2018)0.00
• Conditional Wavegan (2018)4.22
• Unconditional Audio Generation With Generative Adversarial Networks And Cycle Regularization (2020)9.41
• Gansynth: Adversarial Neural Audio Synthesis (2019)0.00
• Waveform Generation For Text-to-speech Synthesis Using Pitch-synchronous Multi-scale Generative Adversarial Networks (2018)8.35
• TFGAN: Time And Frequency Domain Based Generative Adversarial Network For High-fidelity Speech Synthesis (2020)0.00
• A Unified Neural Architecture For Instrumental Audio Tasks (2019)0.00