Perceptual Objective Listening Quality Assessment

Description

Perceptual Objective Listening Quality Assessment (POLQA) is a standardized methodology, defined in ITU-T Recommendation P.863 and adopted by 3GPP, for objectively evaluating the listening quality of speech transmission systems. It is a full-reference algorithm that compares an original reference speech signal with a degraded version received after passing through a network or codec, predicting the Mean Opinion Score (MOS) that human listeners would assign. POLQA works by analyzing both signals in the time-frequency domain, accounting for perceptual factors like loudness, spectral distortion, and temporal artifacts (e.g., delay, packet loss). The algorithm models the human auditory system using a psychoacoustic model, which weights errors based on their perceptual relevance, and includes advanced features for handling modern wideband (WB), super-wideband (SWB), and fullband (FB) audio up to 48 kHz sampling. Key components of POLQA include time alignment to compensate for transmission delays, perceptual transformation to simulate ear processing, and cognitive modeling to derive a MOS score on a scale typically from 1 (bad) to 5 (excellent). In 3GPP networks, POLQA is used to assess voice quality for services like Voice over LTE (VoLTE), Voice over NR (VoNR), and video calls, testing codecs such as AMR-WB, EVS, and Opus under various network conditions (e.g., jitter, packet loss). The process involves injecting test signals into the network, capturing the output, and running POLQA software to generate scores, which help operators optimize network parameters and ensure QoS compliance. POLQA supersedes older algorithms like PESQ (ITU-T P.862), offering improved accuracy for HD voice, noise robustness, and handling of time-warping distortions from variable bitrate codecs. Its role is critical in quality assurance, enabling automated testing without subjective human panels, thus reducing cost and time for network deployment and monitoring. POLQA is integrated into test equipment and network probes, supporting end-to-end and passive monitoring scenarios, and is referenced in 3GPP specs for performance benchmarking.

Purpose & Motivation

POLQA was created to address the limitations of previous speech quality assessment methods, particularly PESQ, which struggled with modern wideband codecs and network impairments found in VoIP and mobile networks. As telecommunications evolved to HD voice services (e.g., VoLTE with AMR-WB) and later to enhanced voice services (EVS) in 4G/5G, there was a need for an objective measurement tool that accurately reflects human perception across broader audio bandwidths and diverse degradation types. The motivation for developing POLQA, standardized by ITU-T in 2011 and adopted by 3GPP from Release 13 onward, was to provide a reliable, standardized metric for evaluating voice quality in all-IP networks, solving problems like inconsistent testing results and inadequate handling of temporal distortions. It enables network operators and equipment vendors to quantify speech performance objectively, ensuring user satisfaction and regulatory compliance without relying on costly subjective listening tests. Historically, subjective MOS tests were time-consuming and variable, while early objective algorithms like PESQ were designed for narrowband circuits and failed with packet loss and jitter common in IP networks. POLQA fills this gap by incorporating advanced psychoacoustic models and supporting super-wideband audio, which is essential for assessing high-definition voice services that deliver superior clarity. Its creation was driven by industry collaboration to standardize testing methodologies, facilitating interoperability and quality benchmarking across global networks. POLQA helps solve quality degradation issues in real-world scenarios, such as background noise, codec transcoding, and network congestion, allowing proactive optimization of voice over LTE and 5G systems.