Exploring Monaural Features for Classification-Based Speech Segregation

By Yuxuan Wang, Kun Han, Deliang Wang. 2013

219 citations recorded by the source corpus.

Subfields: Computational Mechanics, 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 Computational Mechanics

In Signal Processing

Citations in the corpus

Time-Frequency Masking in the Complex Domain for Speech Dereverberation and Denoising (2017, neutral): “A complementary set of features is computed from the given signal [37].”
Binaural Classification for Reverberant Speech Segregation Using Deep Neural Networks (2014, methodology): “GFCC has been shown to be a good single feature in a recent evaluation [39].”
An algorithm to improve speech recognition in noise for hearing-impaired listeners. (2013, background): “Although each of these three feature types can be used to discriminate speech from noise to some degree, the use of all three adds discriminative power (Wang et al., 2013).”
Towards Scaling Up Classification-Based Speech Separation (2013, methodology): “In [46], we have also identified a set of T-F unit-level complementary features that performs very well in matched test conditions.”
On Training Targets for Supervised Speech Separation (2014, methodology): “To focus our study on learning targets, we use a ﬁ xed set of complementary features [34] throughout the experiments.”
Deep clustering: Discriminative embeddings for segmentation and separation (2015, methodology): “The segmentation problem is then solved by segmenting elements into groups or partitions, for example by assigning a group label to each element.”
Complex Ratio Masking for Monaural Speech Separation (2015, background): “A combination of these features has been shown to be effective for speech segregation [16].”
Cocktail Party Processing via Structured Prediction (2012, background): “Wang et al. [12] propose a set of complementary acoustic features that shows further improvements over previous systems.”
Ideal ratio mask estimation using deep neural networks for robust speech recognition (2013, unknown): “We use this group of features as it has been found to work well for IBM estimation [20].”
Investigation of Speech Separation as a Front-End for Noise Robust Speech Recognition (2014, methodology): “We perform T-F masking in the mel-frequency domain unlike some of the recent systems that operate in the gammatone feature domain [17], [42].”
Supervised Speech Separation Based on Deep Learning: An Overview (2017, unknown): “589 A subsequent study [177] explores more monaural features in- 590 cluding mel-frequency cepstral coefficient (MFCC), gamma- 591 tone frequency cepstral coefficient (GFCC) [150], perceptual 592 linear prediction (PLP) [67], and relative spectral transform 593 PLP (RASTA- PLP) [68].”
Speech dereverberation and denoising using complex ratio masks (2017, background): “A complementary feature set is computed from the reverberant (and noisy) signal [18].”
Multi-objective learning and mask-based post-processing for deep neural network based speech enhancement (2017, background): “However the motivation of our work is multi-objective learning with a novel architecture in both input and output layers, which is totally different from the motivation of feature fusion in [33].”
Deep Belief Networks Based Voice Activity Detection (2013, background): “Hence, in real-world applications, researchers still prefer shallow models that are usually easily trained, such as SVM [10], [13], [14], [19], MK-SVM [16], single-hidden-layer-based multilayer perceptron [39], [40], group lasso [41], or Gaussian mixture model [42], [43].”