Perceptual Audio Quality Analysis (POLQA) Test Solution

1. Introduction

In recent years, the forms and application scenarios of online audio and video products have become increasingly diverse, and the development of the epidemic has accelerated the speed of offline to online transformation in many industries. With the normalization of online communication, the technical application of audio and video has become more and more extensive. The participation of major manufacturers such as ZOOM, Tencent, and Alibaba has rapidly improved the quality of this track. In this track, good audio and video quality and good user experience are the key factors to stand out in the fierce competition. In addition to online meetings and other methods, in various calls, no matter what kind of terminal call equipment is used or what kind of network is used, the quality of voice has always been an important indicator for call performance evaluation.

2. Development History

How to systematically evaluate voice quality has been a difficult problem since the invention of the telephone. On August 30, 1996, ITU-T officially launched the P.800: Methods for subjective determination of transmission quality standard. An important indicator was proposed in this standard: MOS (Mean Opinion Score), which is the most widely used subjective evaluation standard today. The MOS scores are 1-5, corresponding to Bad, Poor, Fair, Good, and Excellent.

MOS value rating table

In June 2010, due to the rapid development and application of new transmission technologies and high-order coding methods such as mobile network VoLTE, VoNR, and Internet calling software SKYPE (now MS Teams), the requirements for voice quality assessment methods continued to increase. ITU launched ITU-T P.863 and adopted the joint solution POLQA of OPTICOM, SwissQual and TNO as the new generation of voice quality assessment algorithm. The POLQA algorithm supports a variable input signal level range, a wider signal bandwidth (from narrowband to ultra-wideband) and the latest voice coding technology (Skype, EVS, etc.), and has higher accuracy in evaluating 3G, 4G/LTE and VoIP networks, and expands support for networks that provide high-definition voice services. The POLQA algorithm scores from 1 to 5 points, with 5 points being the highest score. The POLQA algorithm is generally used in online meetings and calls.

3. Introduction to POLQA

1. POLQA is the global standard for benchmarking voice quality in fixed, mobile and IP-based networks. It can be used for voice quality analysis of VoIP, HD Voice, 3G, 4G/VoLTE and 5G networks.

2. POLQA provides a new level of benchmark accuracy and adds important new capabilities for ultra-wideband (HD) and full-band voice signals, as well as support for the latest voice coding and VoIP/VoLTE transmission technologies.

3. POLQA is ideally suited for assessing, optimizing and monitoring voice quality in next-generation networks and provides a standardized, validated Mean Opinion Score (MOS).

4. POLQA generates MOS, speech evaluation results without the need for a large audience, by passing a pre-recorded speech sample through the device under test (DUT) and comparing it to the original speech sample. The algorithm evaluates the difference between the reference audio and the degraded audio and generates the result.

4. Test plan

4.1 Implementation

Our audio analyzer Ax series can fully support the POLQA test. The specific implementation block diagram is as follows:

The signal generation part of our audio analyzer Ax sends the POLQA benchmark test signal to the DUT. The analysis part collects and processes the signal after passing through the DUT. The instrument then automatically compares and analyzes the two sets of signals and gives the MOS value.

4.2 Test Cases

Taking the Bluetooth headset POLQA test as an example, the test solution is mainly divided into uplink test and downlink test:

4.2.1 Uplink test

Note: The uplink test mainly includes the audio analyzer host, electroacoustic test interface, and artificial mouth.

The audio analyzer transmits the test signal to the electroacoustic test interface through a coaxial cable. The electroacoustic test interface transmits the test signal of the audio analyzer to the artificial mouth and supplies power to the artificial mouth at the same time. The artificial mouth plays the test signal to simulate the human sound field. After the microphone of the object under test collects the signal, it is transmitted to the audio analyzer via Bluetooth. The audio analyzer compares the collected sound with the test file to perform POLQA test and give a MOS value.

4.2.2 Uplink test

Note: Downlink test mainly includes the following instruments: audio analyzer host, electroacoustic test interface, and artificial ear test kit.

The audio analyzer transmits the test signal to the object under test via Bluetooth, the artificial ear collects the sound played by the object under test, and the electroacoustic test interface transmits the signal collected by the artificial ear to the audio analyzer while supplying power to the artificial ear. The audio analyzer compares the collected sound with the test file for POLQA testing and gives a MOS value.

5. Introduction to core testing equipment

The A5 audio analyzer sold by Beijing Duwei Technology Co., Ltd. is a professional audio analyzer with high performance, multiple interface types, and multiple test functions. It supports multiple test plug-in expansions and various digital audio interfaces (BT/I²S/HDMI/PDM, etc.), and is the preferred test equipment in the R&D stage of audio and video products such as consumer audio and automotive electronics.

You are welcome to consult and negotiate. The consultation number is: 010-64327909.