Objective Speech Quality Evaluation Research Articles

Imperceptible and Reversible Acoustic Watermarking Based on Modified Integer Discrete Cosine Transform Coefficient Expansion

This paper aims to explore an alternative reversible digital watermarking solution to guarantee the integrity of and detect tampering with data of probative importance. Since the payload for verification is embedded in the contents, algorithms for reversible embedding and extraction, imperceptibility, payload capacity, and computational time are issues to evaluate. Thus, we propose a reversible and imperceptible audio information-hiding algorithm based on modified integer discrete cosine transform (intDCT) coefficient expansion. In this work, the original signal is segmented into fixed-length frames, and then intDCT is applied to each frame to transform signals from the time domain into integer DCT coefficients. Expansion is applied to DCT coefficients at a higher frequency to reserve hiding capacity. Objective evaluation of speech quality is conducted using listening quality objective mean opinion (MOS-LQO) and the segmental signal-to-noise ratio (segSNR). The audio quality of different frame lengths and capacities is evaluated. Averages of 4.41 for MOS-LQO and 23.314 [dB] for segSNR for 112 ITU-T test signals were obtained with a capacity of 8000 bps, which assured imperceptibility with the sufficient capacity of the proposed method. This shows comparable audio quality to conventional work based on Linear Predictive Coding (LPC) regarding MOS-LQO. However, all segSNR scores of the proposed method have comparable or better performance in the time domain. Additionally, comparing histograms of the normalized maximum absolute value of stego data shows a lower possibility of overflow than the LPC method. A computational cost, including hiding and transforming, is an average of 4.884 s to process a 10 s audio clip. Blind tampering detection without the original data is achieved by the proposed embedding and extraction method.

Non-Intrusive Air Traffic Control Speech Quality Assessment with ResNet-BiLSTM

In the current field of air traffic control speech, there is a lack of effective objective speech quality evaluation methods. This paper proposes a new network framework based on ResNet–BiLSTM to address this issue. Firstly, the mel-spectrogram of the speech signal is segmented using the sliding window technique. Next, a preceding feature extractor composed of convolutional and pooling layers is employed to extract shallow features from the mel-spectrogram segment. Then, ResNet is utilized to extract spatial features from the shallow features, while BiLSTM is used to extract temporal features, and these features are horizontally concatenated. Finally, based on the concatenated spatiotemporal features, the final speech quality score is computed using fully connected layers. We conduct experiments on the air traffic control speech database and compare the objective scoring results with the subjective scoring results. The experimental results demonstrate that the proposed method has a high correlation with the mean opinion score (MOS) of air traffic control speech.

A Smart Control System for the Oil Industry Using Text-to-Speech Synthesis Based on IIoT

Oil refineries have high operating expenses and are often exposed to increased asset integrity risks and functional failure. Real-time monitoring of their operations has always been critical to ensuring safety and efficiency. We proposed a novel Industrial Internet of Things (IIoT) design that employs a text-to-speech synthesizer (TTS) based on neural networks to build an intelligent extension control system. We enhanced a TTS model to achieve high inference speed by employing HiFi-GAN V3 vocoder in the acoustic model FastSpeech 2. We experimented with our system on a low resources-embedded system in a real-time environment. Moreover, we customized the TTS model to generate two target speakers (female and male) using a small dataset. We performed an ablation analysis by conducting experiments to evaluate the performance of our design (IoT connectivity, memory usage, inference speed, and output speech quality). The results demonstrated that our system Real-Time Factor (RTF) is 6.4 (without deploying the cache mechanism, which is a technique to call the previously synthesized speech sentences in our system memory). Using the cache mechanism, our proposed model successfully runs on a low-resource computational device with real-time speed (RTF equals 0.16, 0.19, and 0.29 when the memory has 250, 500, and 1000 WAV files, respectively). Additionally, applying the cache mechanism has reduced memory usage percentage from 16.3% (for synthesizing a sentence of ten seconds) to 6.3%. Furthermore, according to the objective speech quality evaluation, our TTS model is superior to the baseline TTS model.

A Complex Algorithm for Objective Evaluation of the Decoded Speech Signal Quality Under the Action of Acoustic Interference

The paper presents algorithms for objective evaluation of speech quality are considered, based on the measurement of dynamic and static characteristics of speech signals at the source codec output. The functional scheme of carrying out of experimental researches is proved. The results of the analysis of the correlation of objective and subjective evaluation of speech quality are given. Modifications of the objective quality assessment are proposed on the basis of the correlation of the excitation of the MESC spectrum and modification of the exponent on the basis of the calculation of the sensation function of the spectral dynamics of MFOSD. The algorithm of regression curve formation is proposed, which allows to perform the transformation of objective evaluation to the scale of subjective evaluation of speech quality. Based on the use of the most accurate modifications of the speech quality assessment indicators for reconstructed speech signals, a complex algorithm for objective hardware evaluation of the speech quality is proposed when the broadband and low-frequency stationary and nonstationary acoustic waves are applied to the microphone. It is shown that the use of a complex algorithm makes it possible to obtain an objective evaluation of the quality of speech according to GOST R 50840-95 with an average error of no more than 0.35 points for signal-to-noise ratios of 30 dB to -10 dB.

Objective Speech Quality Evaluation Based on Matlab

The speech quality is one of the most important indexes to evaluate the communication system. The objective evaluation is the last goal that people have been pursuing. The most efficient evaluation methods based on the spectrum distortion measure was analysed. The objective evaluation method based on wavelet spectrum distortion measure was simulated on the Matlab simulation environment. The simulation results show that Matlab was very suit to analyse the speech quality. The objective evaluation rules are found out on Matlab and the objective evaluation method was achieved.

On the identification of relevant degradation indicators in super wideband listening quality assessment models

Recently, new objective speech quality evaluation methods, designed and adapted to new high voice quality contexts, have been developed. One interest of these methods is that they integrate voice quality perceptual dimensions reflecting the effects of frequency–response distortions, discontinuities, noise and/or speech level deviations respectively. This makes it possible to use these methods also to provide diagnostic information about specific aspects of the transmission systems’ quality, as perceived by end-users. In this paper, we present and analyze in depth two of these approaches namely POLQA (Perceived Objective Listening Quality Assessment) and DIAL (Diagnostic Instrumental Assessment of Listening quality), in terms of quality degradation indicators related to the perceptual dimensions these models could embed. The main goal of our work is to find and propose the most robust quality degradation indicators to reliably characterize the impact of degradations relative to the perceptual dimensions described above and to identify the underlying technical causes in super wideband telephone communications [50, 14000]Hz. To do so, the first step of our study was to identify in both models the correspondence between perceptual dimensions and quality degradation indicators. Such indicators could be either present in the model itself or derived from our own investigation of the model. In a second step, we analyzed the performance and robustness of the identified quality degradation indicators on speech samples only impaired by one degradation (representative of one perceptual dimension) at a time. This study highlighted the reliability of some of the quality degradation indicators embedded in the two models under study and stood for a first step in the evaluation of performance of these indicators to quantify the degradation for which they were designed.

A Novel Recurrent Generalized Congruence Neural Network Measure for Objective Speech Quality Evaluation

A new system model for objective speech quality evaluation based on the improved recurrent generalized congruence neural network (RGCNN/OSQE) is proposed. The performance of the RGCNN model is compared with the most commonly used RBFNN (radial basis function neural network) model in objective speech quality evaluation. Comparison results show that the RGCNN model has higher correlation coefficient, less deviation, and saves about half training time, i.e., the RGCNN model has obvious advantages over the RBFNN model. Therefore, the novel RGCNN model for objective speech quality evaluation is feasible and effective.

Objective Evaluation System of Speech Quality Based HMM

It is of great significance that objective evaluation of speech quality in language self-learning. Aimed at the objective evaluation process of speech quality the thesis presented to build HMM model based speech feature MFCC of human auditory model, and to use different indicators to achieve the evaluation of phoneme, word and sentence. Effective solution was proposed for such issues as the initial model to select more sequence of observations to train HMM as well as data underflow encountered during practical application of HMM algorithm. According to some speech experts, the evaluation results achieved by speech quality evaluation system were objective.

ON A METHOD FOR SOLUTION OF THE PROBLEM ON DETERMINATION OF OPTIMAL CONTROL PARAMETERS OF A LOW-RATE VOICE INFORMATION COMPRESSION SYSTEM

A method for direct solution of the problem on determination of optimal control parameters of a voice information compression system in the process of its development is discussed. It is proposed to use a method of measuring the cepstral distance between the original voice signals and voice signals transformed by the compression system for a rapid speech quality evaluation. A general functional layout of the device for a rapid objective speech quality evaluation by the method of cepstral distance measurement is described.

A Multiresolution Model of Auditory Excitation Pattern and Its Application to Objective Evaluation of Perceived Speech Quality

This paper proposes a multiresolution model of auditory excitation pattern and applies it to the problem of objective evaluation of subjective wideband speech quality. The model uses wavelet packet transform for time-frequency decomposition of the input signal. The selection of the wavelet packet tree is based on an optimality criterion formulated to minimize a cost function based on the critical band structure. The models of the different auditory phenomena are reformulated for the multiresolution framework. This includes the proposition of duration dependent outer and middle ear weighting, multiresolution spectral spreading, and multiresolution temporal smearing. As an application, the excitation pattern is used to define an objective measure of auditory distortion of a distorted speech signal compared to the undistorted one. The performance of this objective measure is evaluated with a database of various kinds of NOISEX-92 degraded wideband speech signals in predicting the subjective mean opinion score (MOS) and is compared with the fast Fourier transform (FFT)-based ITU-T PESQ P.862.2 algorithm. The proposed measure is found to achieve comparable correlation between subjective MOS and objective MOS as PESQ P.862.2, with a trend suggesting better correlation for the nonstationary degradations compared to the stationary ones. Further refinement of the measure for distortion types other than additive noise is anticipated

Psychoacoustically motivated objective speech quality evaluation procedures, PSQM, and improvements

PSQM (Perceptual Speech Quality Measure), measuring speech quality objectively, has been standardized by ITU-T as recommendation P.861. PSQM characterizes the perception of the (degraded) output speech signal of the system in comparison to the (ideal) input speech. A perceptual model is used that maps input and output signals onto psychophysical representations using psychophysical equivalents of frequency (Bark) and intensity (compressed Sone). The quality of the device under test is determined with a simple cognitive mapping from the differences in the psychophysical representation to the perceived speech quality in terms of Mean Opinion Scores (MOS). Within an ITU benchmark testing a limited set of unknown codec distortions the PSQM showed high correlations (around 0.97) between subjectively perceived and objectively measured speech quality. When applying PSQM to a wide variety of real world distortions two major limitations show up: First, dynamic time warping effects, as they will be found, e.g., in Internet telephony, cause a degradation in correlation. Second, the perceptual model that is used within the PSQM method is too simple to account for a wide variety of distortions. This presentation discusses extensions to the PSQM method that allows application to real world distortions.