Monaural speech segregation based on pitch tracking and amplitude modulation

By Guoning Hu, Deliang Wang. IEEE Transactions on Neural Networks, 2004

458 citations recorded by the source corpus.

View Paper: https://www.semanticscholar.org/paper/53ed39106dec6486e2560d022b5ccb337066f9a0

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

1. Yuxuan Wang
2. Deliang Wang
3. Kun Han
4. Yi-jiao Jiang
5. Chao-Ling Hsu
6. Runsheng Liu
7. Zhaozhang Jin
8. Yipeng Li

Citations in the corpus

• On the optimality of ideal binary time-frequency masks (2008, methodology): “Several CASA algorithms that directly estimate the IBM [5, 4] have produced good results in speech separation.”
• Monaural Sound Source Separation by Nonnegative Matrix Factorization With Temporal Continuity and Sparseness Criteria (2007, background): “extract only the most prominent source [1], [2], or which utilize prior information of the source signals [3], [4].”
• On Ideal Binary Mask As the Computational Goal of Auditory Scene Analysis (2005, background): “The model of Hu and Wang (2004) produces substantially better performance than previous models, especially in the high-frequency range.”
• Exploring Monaural Features for Classification-Based Speech Segregation (2013, methodology): “used as the ground truth signal [15], [17], as the IBM represents the ground truth of classification.”
• Supervised Speech Separation Based on Deep Learning: An Overview (2017, unknown): “In addition to the traditional SNR, 360 speech distortion (loss) and noise residue in a separated signal 361 can be individually measured [77], [113].”
• A Tandem Algorithm for Pitch Estimation and Voiced Speech Segregation (2010, methodology): “We have suggested that the IBM is a reasonable goal for CASA [16], [37], and it has since been used as a measure of ceiling performance for speech separation [26], [31], [32].”
• A Supervised Learning Approach to Monaural Segregation of Reverberant Speech (2009, methodology): “The next stage is to label each of the T-F units using MLP. Previous studies [24,45] treat a single pitch-based feature as the grouping cue and rely on thresholding for unit labeling.”
• Cocktail Party Processing via Structured Prediction (2012, methodology): “This way of computing SNRs is commonly adopted in the literature [4, 25], as the IBM represents the ground truth of classification.”
• Binaural Classification for Reverberant Speech Segregation Using Deep Neural Networks (2014, methodology): “The resynthesized speech from the IBM is used as the ground truth since the IBM is the ground truth of DNN classification [17]:”
• Dynamic Precedence Effect Modeling for Source Separation in Reverberant Environments (2010, methodology): “lized signal-to-noise ratio metric proposed by Hu and Wang [16] was employed.”
• An Overview of Speaker Identification: Accuracy and Robustness Issues (2011, background): “Reliability decisions are achieved via monaural voiced speech segregation obtained from accurate pitch contour tracking [94].”
• On the Improvement of Singing Voice Separation for Monaural Recordings Using the MIR-1K Dataset (2010, methodology): “In this block, we perform time–frequency decomposition by using the same settings as in [5] and [13].”