D3Net: Densely connected multidilated DenseNet for music source separation
By Naoya Takahashi, Yuki Mitsufuji. 2020
85 citations recorded by the source corpus.
Music source separation involves a large input field to model a long-term dependence of an audio signal. Previous convolutional neural network (CNN) -based approaches address the large input field modeling using sequentially down- and up-sampling feature maps or dilated convolution. In this paper, we claim the importance of a rapid growth of a receptive field and a simultaneous modeling of multi-resolution data in a single convolution layer, and propose a novel CNN architecture called densely connected dilated DenseNet (D3Net). D3Net involves a novel multi-dilated convolution that has different dilation factors in a single layer to model different resolutions simultaneously. By combining the multi-dilated convolution with DenseNet architecture, D3Net avoids the aliasing problem that exists when we naively incorporate the dilated convolution in DenseNet. Experimental results on MUSDB18 dataset show that D3Net achieves state-of-the-art performance with an average signal to distortion ratio (SDR) of 6.01 dB.
Subfields: Signal Processing
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Signal Processing
- Alexandre D'efossez
- Francis R. Bach
- Francisco Massa
- Nicolas Usunier
- Simon Rouard
- L. Eon Bottou
- Yossi Adi
- Gabriel Synnaeve
Citations in the corpus
- Music Source Separation in the Waveform Domain (2019, methodology): “We also compare to the more recent options like the D3Net [Takahashi and Mitsufuji, 2020] model, which used dilated convolutions with dense connections, as well as the Spleeter model [Hennequin et al.”
- Hybrid Transformers for Music Source Separation (2022, methodology): “The former includes models like Open-Unmix [11], a biLSTM with fully connected that predicts a mask on the input spectrogram or D3Net [12] which uses dilated convolutional blocks with dense connections.”