Neural Audio Synthesis of Musical Notes with WaveNet Autoencoders
This work addresses the challenge of high-quality audio synthesis for music applications, offering incremental improvements in generative modeling for audio.
The paper tackles the problem of generating realistic musical audio by introducing a WaveNet-style autoencoder model and a large-scale dataset called NSynth, resulting in improved performance over a baseline and the ability to create new sounds through timbre interpolation.
Generative models in vision have seen rapid progress due to algorithmic improvements and the availability of high-quality image datasets. In this paper, we offer contributions in both these areas to enable similar progress in audio modeling. First, we detail a powerful new WaveNet-style autoencoder model that conditions an autoregressive decoder on temporal codes learned from the raw audio waveform. Second, we introduce NSynth, a large-scale and high-quality dataset of musical notes that is an order of magnitude larger than comparable public datasets. Using NSynth, we demonstrate improved qualitative and quantitative performance of the WaveNet autoencoder over a well-tuned spectral autoencoder baseline. Finally, we show that the model learns a manifold of embeddings that allows for morphing between instruments, meaningfully interpolating in timbre to create new types of sounds that are realistic and expressive.