Joint Dictionary Learning-Based Non-Negative Matrix Factorization for Voice Conversion to Improve Speech Intelligibility After Oral Surgery

Abstract

This paper focuses on machine learning based voice conversion (VC) techniques for improving the speech intelligibility of surgical patients who have had parts of their articulators removed. Because of the removal of parts of the articulator, a patient's speech may be distorted and difficult to understand. To overcome this problem, VC methods can be applied to convert the distorted speech such that it is clear and more intelligible. To design an effective VC method, two key points must be considered: 1) the amount of training data may be limited (because speaking for a long time is usually difficult for postoperative patients); 2) rapid conversion is desirable (for better communication). We propose a novel joint dictionary learning based non-negative matrix factorization (JD-NMF) algorithm. Compared to conventional VC techniques, JD-NMF can perform VC efficiently and effectively with only a small amount of training data. The experimental results demonstrate that the proposed JD-NMF method not only achieves notably higher short-time objective intelligibility (STOI) scores (a standardized objective intelligibility evaluation metric) than those obtained using the original unconverted speech but is also significantly more efficient and effective than a conventional exemplar-based NMF VC method. The proposed JD-NMF method may outperform the state-of-the-art exemplar-based NMF VC method in terms of STOI scores under the desired scenario. We confirmed the advantages of the proposed joint training criterion for the NMF-based VC. Moreover, we verified that the proposed JD-NMF can effectively improve the speech intelligibility scores of oral surgery patients. This paper focuses on machine learning based voice conversion (VC) techniques for improving the speech intelligibility of surgical patients who have had parts of their articulators removed. Because of the removal of parts of the articulator, a patient's speech may be distorted and difficult to understand. To overcome this problem, VC methods can be applied to convert the distorted speech such that it is clear and more intelligible. To design an effective VC method, two key points must be considered: 1) the amount of training data may be limited (because speaking for a long time is usually difficult for postoperative patients); 2) rapid conversion is desirable (for better communication). We propose a novel joint dictionary learning based non-negative matrix factorization (JD-NMF) algorithm. Compared to conventional VC techniques, JD-NMF can perform VC efficiently and effectively with only a small amount of training data. The experimental results demonstrate that the proposed JD-NMF method not only achieves notably higher short-time objective intelligibility (STOI) scores (a standardized objective intelligibility evaluation metric) than those obtained using the original unconverted speech but is also significantly more efficient and effective than a conventional exemplar-based NMF VC method. The proposed JD-NMF method may outperform the state-of-the-art exemplar-based NM...

Speech Classification using Logical ART Deep Mechanism of Machine Learning

“…The computational modeling is mechanized based on an effective training-classification paradigm. The experimental analysis performed clearly shows that the system can effectively recognize and classify the speech signal prime entities as compared to existing baselines [13]. There exist several applications which process lower operational features to synthesize speech signals.…”

Locality Preserved Joint Nonnegative Matrix Factorization for Speech Emotion Recognition

“…This method used two types of features, multiple-frame lowresolution exemplars and single-frame high-resolution exemplars, to estimate the activation matrix simultaneously. Fu et al [14] introduced joint dictionary learning-based NMF (JD-NMF) used to improve patients' speech intelligibility after oral surgery. The JD-NMF was designed to learn the source and the target dictionaries simultaneously from the common activation matrix.…”

Nonparallel Dictionary-Based Voice Conversion Using Variational Autoencoder with Modulation-Spectrum-Constrained Training

“…In another method, the matrices A and Z are learned from the training data by alternately updating one matrix while keeping the other matrix fixed. The size of the constructed dictionary using this method is significantly reduced relative to that in the exemplar-based NMF method, resulting in improved online conversion efficiency [3].…”

“…The baseline system is the NMF-based voice conversion system using parallel data described in [3]. The dictionaries have r = 100 bases.…”