Improving Singing Voice Separation with the Wave-U-Net Using Minimum Hyperspherical Energy

Perez-Lapillo, Joaquin; Galkin, A. Yu.; Weyde, Tillman

doi:10.1109/icassp40776.2020.9053424

Cited by 11 publications

(4 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Cohen-Hadria et al [169] investigated data augmentation techniques such as pitch-shifting and time-scaling on publicly available smaller datasets and compared the performance of U-Net and Wave-U-Net models. In order to avoid overfitting, besides using data augmentation, Perez-Lapillo et al [170] employed an additional regularization term in the loss function, called the minimum hyperspherical energy for the Wave-U-Net architecture, where the diversity of neurons is promoted by minimizing the hyperspherical energy of the neurons in each layer. Since convolutional encoder-decoder frameworks are sensitive to the sound level of the input, Lin et al [171] showed that a combination of data augmentation, frame normalization, and zero-mean convolution makes the network sound-level invariant.…”

Section: Singing Voice Separationmentioning

confidence: 99%

Deep Learning Approaches in Topics of Singing Information Processing

Gupta

Goto

2022

IEEE/ACM Trans. Audio Speech Lang. Process.

View full text Add to dashboard Cite

Singing, the vocal production of musical tones, is one of the most important elements of music. Addressing the needs of real-world applications, the study of technologies related to singing voices has become an increasingly active area of research. In this paper, we provide a comprehensive overview of the recent developments in the field of singing information processing, specifically in the topics of singing skill evaluation, singing voice synthesis, singing voice separation, and lyrics synchronization and transcription. We will especially focus on deep learning approaches including modern representation learning techniques for singing voices. We will also provide an overview of contributions in public datasets for singing voice research.

show abstract

Section: Singing Voice Separationmentioning

confidence: 99%

Deep Learning Approaches in Topics of Singing Information Processing

Gupta

Goto

2022

IEEE/ACM Trans. Audio Speech Lang. Process.

View full text Add to dashboard Cite

show abstract

“…In this study, the 1DCNN model was trained with a batch size of 128 and a maximum of 500 epochs using the Adam optimizer [34] to minimize the multitasking loss generated by calculating the sum of the product of each task-specific loss and the loss weighting. Besides, a dropout rate of 0.2 and early stopping patience of 15 (epochs without improvement of test loss) [35] were also implemented to prevent overfitting.…”

Section: One-dimension Convolutional Neural Network (1dcnn)mentioning

confidence: 99%

The Simultaneous Prediction of Soil Properties and Vegetation Coverage from Vis-NIR Hyperspectral Data with a One-Dimensional Convolutional Neural Network: A Laboratory Simulation Study

et al. 2022

View full text Add to dashboard Cite

Remote sensing of land surface mostly obtains a mixture of spectral information of soil and vegetation. It is thus of great value if soil and vegetation information can be acquired simultaneously from one model. In this study, we designed a laboratory experiment to simulate land surface compositions, including various soil types with varying soil moisture and vegetation coverage. A model of a one-dimensional convolutional neural network (1DCNN) was established to simultaneously estimate soil properties (organic matter, soil moisture, clay, and sand) and vegetation coverage based on the hyperspectral data measured in the experiment. The results showed that the model achieved excellent predictions for soil properties (R2 = 0.88–0.91, RPIQ = 4.01–5.78) and vegetation coverage (R2 = 0.95, RPIQ = 7.75). Compared with the partial least squares regression (PLSR), the prediction accuracy of 1DCNN improved 42.20%, 45.82%, 43.32%, and 36.46% in terms of the root-mean-squared error (RMSE) for predicting soil organic matter, sand, clay, and soil moisture, respectively. The improvement might be caused by the fact that the spectral preprocessing and spectral features useful for predicting soil properties were successfully identified in the 1DCNN model. For the prediction of vegetation coverage, although the prediction accuracy by 1DCNN was excellent, its performance (R2 = 0.95, RPIQ = 7.75, RMSE = 3.92%) was lower than the PLSR model (R2 = 0.98, RPIQ = 12.57, RMSE = 2.41%). These results indicate that 1DCNN can simultaneously predict soil properties and vegetation coverage. However, the factors such as surface roughness and vegetation type that could affect the prediction accuracy should be investigated in the future.

show abstract

“…To alleviate these difficulties, Lin et al [41] propose the compressive MHE (CoMHE) as a more effective regularization to minimize hyperspherical energy for neural networks. Following [18], [41], Perez-Lapillo et al [42] and Shah et al [43] improve voice separation by applying MHE to Wave-U-Net and timefrequency domain networks, respectively. MHE has wide applications in image recognition [39], [44], [45], face recognition [36], [18], [46], speaker verification [47], adversarial robustness [48], few-shot learning [49], [50], etc.…”

Section: Related Workmentioning

confidence: 99%

Data-Efficient Learning via Minimizing Hyperspherical Energy

Cao¹,

Liu²,

Tsang³

2022

Preprint

View full text Add to dashboard Cite

Deep learning on large-scale data is dominant nowadays. The unprecedented scale of data has been arguably one of the most important driving forces for the success of deep learning. However, there still exist scenarios where collecting data or labels could be extremely expensive, e.g., medical imaging and robotics. To fill up this gap, this paper considers the problem of data-efficient learning from scratch using a small amount of representative data. First, we characterize this problem by active learning on homeomorphic tubes of spherical manifolds. This naturally generates feasible hypothesis class. With homologous topological properties, we identify an important connection -finding tube manifolds is equivalent to minimizing hyperspherical energy (MHE) in physical geometry. Inspired by this connection, we propose a MHE-based active learning (MHEAL) algorithm, and provide comprehensive theoretical guarantees for MHEAL, covering convergence and generalization analysis. Finally, we demonstrate the empirical performance of MHEAL in a wide range of applications on data-efficient learning, including deep clustering, distribution matching, version space sampling and deep active learning.

show abstract

Improving Singing Voice Separation with the Wave-U-Net Using Minimum Hyperspherical Energy

Cited by 11 publications

References 17 publications

Deep Learning Approaches in Topics of Singing Information Processing

Deep Learning Approaches in Topics of Singing Information Processing

The Simultaneous Prediction of Soil Properties and Vegetation Coverage from Vis-NIR Hyperspectral Data with a One-Dimensional Convolutional Neural Network: A Laboratory Simulation Study

Data-Efficient Learning via Minimizing Hyperspherical Energy

Contact Info

Product

Resources

About