Towards Improved Objective Perceptual Audio Quality Assessment -- Part 1: A Novel Data-Driven Cognitive Model

TL;DR

A novel machine learning approach that uses subjective data to model cognitive aspects of audio quality perception is introduced, allowing potential extensions and improvements to multi-dimensional quality measurement without complete retraining.

Abstract

Efficient audio quality assessment is vital for streamlining audio codec development. Objective assessment tools have been developed over time to algorithmically predict quality ratings from subjective assessments, the gold standard for quality judgment. Many of these tools use perceptual auditory models to extract audio features that are mapped to a basic audio quality score prediction using machine learning algorithms and subjective scores as training data. However, existing tools struggle with generalization in quality prediction, especially when faced with unknown signal and distortion types. This is particularly evident in the presence of signals coded using non-waveform-preserving parametric techniques. Addressing these challenges, this two-part work proposes extensions to the Perceptual Evaluation of Audio Quality (PEAQ - ITU-R BS.1387-1) recommendation. Part 1 focuses on increasing generalization, while Part 2 targets accurate spatial audio quality measurement in audio coding. To enhance prediction generalization, this paper (Part 1) introduces a novel machine learning approach that uses subjective data to model cognitive aspects of audio quality perception. The proposed method models the perceived severity of audible distortions by adaptively weighting different distortion metrics. The weights are determined using an interaction cost function that captures relationships between distortion salience and cognitive effects. Compared to other machine learning methods and established tools, the proposed architecture achieves higher prediction accuracy on large databases of previously unseen subjective quality scores. The perceptually-motivated model offers a more manageable alternative to general-purpose machine learning algorithms, allowing potential extensions and improvements to multi-dimensional quality measurement without complete retraining.

Figures (10)

• Figure 1: Block diagram of the proposed objective quality assessment system. The perceptual model corresponds to that used in the PEAQ method PEAQ. The feature extraction stage contemplates the existing PEAQ features (advanced version), as distortion metrics in the time-pitch-loudness domain through the excitation patterns. The proposed extension uses additional features for the effect size measurement of different cognitive phenomena. The novel cognitive salience model structure constitutes an explicit interaction model of cognitive effects and distortion metrics.
• Figure 2: Block diagram for the interaction selection and optimization procedure.
• Figure 3: Basis Functions mapping individual DM outputs to quality scores estimated with the procedure described in Section \ref{['sec:BF']}.
• Figure 4: Mean Effect Size for a CEM (speech probability) and DM salience (linear distortions) for the signals in the USAC Verification Test Database 1. The DPW transformation increases DM/CEM covariance given by $\mathcal{C}$ as calculated by Equation \ref{['eq:costFunction']}.
• Figure 5: An illustrative CSM model output for an audio item of the USAC Verification Test 1 database. The signal is composed of an orchestral music excerpt followed by female speech, around frame 640. The CSM model steers the objective score output in time towards the salient DM basis function output according to the probability of the signal being speech-like. Top plot: reference signal loudness. Second-to-top: probability of speech-like signal. Middle plot: noise loudness BF outputs and corresponding DPW. Second-to-bottom plot: missing component loudness BF output and corresponding DPW. Bottom Plot: CSM output over time and the mean MUSHRA subjective score for the analyzed item. Temporal granularity is $t_s = 0.1 s$ per frame.