Tag Archive for: module

The New POLQA Module in SoundCheck

The optional POLQA (Perceptual Objective Listening Quality Analysis) module brings Opticom’s POLQA 3 algorithm, as specified in the ITU-T P.863 standard, right into SoundCheck (Version 20 and later). POLQA Analysis is a fast and cost-effective alternative to a panel of human listeners for perceptual measurements of speech degradation in communications applications ranging from telephones to smart devices. The quality of the speech signal is evaluated by comparing the degraded signal to the corresponding reference signal to determine the Mean Opinion Score (MOS). This end-to-end analysis may be implemented using either acoustic, digital, or electrical signals, and with either narrowband (300-3400Hz) or full band (10-20000Hz) signal analysis.

The New POLQA Module in SoundCheck

Ready to try for yourself?

If you are ready to add the new POLQA module to your SoundCheck system, contact your local sales engineer or sales@listeninc.com with any questions.

Learn more about POLQA and SoundCheck

Read on about all POLQA algorithm functionality, licensing, applications, and more.

 

Video Script: Using POLQA Analysis in SoundCheck

Opticom’s POLQA version 3 algorithm as specified in the ITU-T P.863 standard is integrated directly into SoundCheck. It operates just like any other post-processing step. Within soundcheck you can configure the reference and response waveforms, set test parameters and define outputs through the familiar soundcheck interface. The POLQA algorithm is called from within soundcheck and returns results as graphs and tables like any other analysis. It can of course be used within sequences to accelerate and simplify test procedures and can also be used to process imported data using batch file processing.

POLQA is a valuable addition for anyone testing speech quality and communications devices. Some common applications of this test might include assessing the impact of a noise reduction algorithm on an audio communication system, evaluating the perceived audio degradation in a Bluetooth connection due to packet loss, analyzing distortions introduced in the audio path, or quantifying the impact of various speech codecs on perceived speech quality. Let’s look at a few possible  POLQA setups.

Here is an example of a setup we can use for characterizing the perceived audio degradation at a Bluetooth connection due to packet loss. The mouth simulator plays back a reference WAV file that is recorded through the Bluetooth microphone, the recorded degraded responses are compared to the original reference WAV file, and the mean opinion score or MOS is returned.

Here’s an example of another possible  POLQA test. In this setup, an end-to-end call can be established, for example a call can be made to a conference phone which is the device under test, and the signal on the receiving end is recorded and analyzed for speech quality. Let’s look at a simple  POLQA SoundCheck sequence.

This sequence is similar to the conference phone setup I just displayed. The sequence uses the batch processing capability introduced into SoundCheck 20 to access pairs of wav files for  POLQA analysis. These wav files are temporarily loaded into the memory list and passed through the  POLQA post-processing step for analysis. This loop recalls and analyzes four sets of wav files, and reports the POLQA mean opinion score on a scale of one to five one being poor, five being best, and a score of around four considered very good.

The advantage of batch processing is that I can conceivably test hundreds or thousands of pairs of wav files without loading all the files into a sequence at once, efficiently managing the sequence size, memory, and performance. Let’s look at the  POLQA post processing step. The step compares the recall to reference and response files; the drop list allows you to select the measurement type – acoustic, digital or voltage. This drop-down shows the operating modes options including full band, narrow band, and super wide band including the latest version 3 algorithm. It also supports Opticom version 2.4 if you need to correlate results with a lab that uses the Legacy 2.4 algorithm. In addition to the MOS, we can choose additional outputs including ASR, attenuation, delay, MOS per frame, pitch, R-factor and SNR.

Since the sequence is using recall wav files you won’t hear the speech files when I run the sequence but I’ll play back the reference WAV file through my signal generator so you can hear what’s being analyzed. I’ll run the sequence and note the highlighted steps in the sequence editor particularly through the loop which will run four times before the final display steps. There are two tabs in my display step. As you watch the sequence run this display set shows the active iteration of the four wav files being analyzed. The top display shows the reference WAV file, and the bottom displays the reference wav file along with the response file superimposed. In the second display step we see the mean opinion scores, or MOS, of the four wav files and additional outputs below. These are displayed in tables and we can see the MOS scores are 3.4, 3.98, 3.4 and 3.98. As you can see, SoundCheck’s integration of  POLQA is fast and easy and an essential addition for characterizing speech quality.