Useful information about using SoundCheck from our team

New Transient Distortion Measurement Algorithm

At Listen, we’re at the forefront of audio measurement research, and we’re always looking to improve on existing audio measurement techniques, even our own! Loose Particle detection has been a valuable production metric for analyzing transient distortion since we launched it in 2004. This new iteration uses our own original research to improve accuracy and reliability, show a clear correlation to audibility, and simplify limit setting. In addition to production line testing of speakers, headphones, drivers and other devices, it is also valuable for automotive Buzz, Squeak and rattle (BSR) measurements and measuring rattling components such as keys and buttons on a variety of devices. All is explained in the short video below.

Transient Distortion Detection Launch Video

Watch our launch video (broadcast date 05/25/2023) for the full details.

 

Ready to Measure your Transient Distortion?

If you have SoundCheck 21, you already have this new algorithm! Download our free complete end of line test sequence with enhanced Loose Particles to start using it right away.

If you don’t have SoundCheck 21, but have an older version, you can send us your recorded waveforms and we’ll analyze them for you and send you the results. Please contact sales@listeninc.com for a test sequence to record the waveforms.

No SoundCheck system at all? No problem! You can send us your speakers and we’ll test them for you. Or we may be able to arrange a system loan. Contact your sales engineer at sales@listeninc.com for more information.

Prefer to Read About It?

We know not everyone has 15 mins free to watch a video, although its hard to beat the benefits of a proper demonstration. So here’s a brief summary of the information about this new algorithm that is presented in the video:

What it does

The new algorithm measures transient distortion caused by loose particles that may become trapped in a device during manufacture and create an unpleasant sound when they vibrate in the finished product. This is measured in the time domain rather than the frequency domain as these artifacts appear randomly over time, and not periodically in the same way that harmonic distortion artifacts are presented. This algorithm has a couple of unique features:

1) It measures transient distortion separately from harmonic distortion which gives deeper insight into the failure mode and accelerates troubleshooting, particularly on the production line.

2) It’s easy to correlate with audibility as the algorithm removes the stimulus waveform to allow the user to listen to just the distortion artifacts. As well as enabling the user to truly understand how measured results correlate to listening, this also facilitates limit setting.

3) It’s reliable even in the presence of transient background noise since it relies on a cumulative event count rather than a single event triggering a fail. Limit setting is simple as it is not frequency dependent

 

Applications

  • Production line driver test
  • Production line finished product test (rattling buttons, keys, grills, and other components)
  • Buzz, Squeak and Rattle (BSR) / Impulsive distortion measurements in cars

 

How it works

Like Listen’s original 2004 transient distortion detection algorithm, enhanced Loose Particles relies on a time domain analysis of the waveform. However, rather than simply filtering and counting transients, it removed the stimulus, performs a time domain analysis of the remaining transients, then applies a unique Prominence calculation to evaluate each transient in the context of the surrounding waveform. A threshold is applied, and the number of transient events above that threshold are counted. The event count indicates whether a device has a transient distortion problem. Transient events caused by speaker manufacturing issues tend to repeat many times as particles bounce around in side the speaker, or components vibrate. In contrast, background noise events occur infrequently during the duration of the measurement. The threshold is set based on audibility, and the Loose Particle count limit can be set according to the environment to fine-tune the algorithm for the specific operating conditions.

 

Graphic demonstrating how the Transient Distortion algorithm (Loose Particles) works

4 Stages of the enhanced Loose Particles algorithm for transient distortion measurement: Response waveform, Loose Particle waveform, Prominence and enhanced Loose Particles.

 

 

For additional background, comparison to other methods, such as Crest Factor Analysis, and live demonstrations, please check out the video above.

 

Learn More

AES75 (M-Noise) Measurement of Max SPL for Loudspeakers

The new AES75-2022 standard for Measuring Loudspeaker Maximum Linear Sound Levels Using Noise is a complex test process which uses the M-Noise test signal developed by Meyer Sound to measure the maximum linear sound levels of a loudspeaker system or driver. This test signal was specifically developed to emulate the dynamic characteristics of music.

Implementing the standard manually relies on an operator’s subjective judgment of real time spectrum analyzer data, and is labor intensive since it requires multiple iterations of a measurement. Our free SoundCheck test sequence fully automates the entire measurement, using calculated results to drive subsequent steps in the procedure. This removes subjectivity, increasing reliability, and saving time. This short video introduces the test sequence with a short demonstration.

Video Demo

Try it Yourself

Would you like to try this yourself? If you already have SoundCheck, you can download the AES75 (M-Noise) test sequence. Please note that you will need the waveform filter (part # 2032) and transfer function (part # 2021) modules installed on your SoundCheck system. You can download the AES75 Standard from the Audio Engineering Society website.

 

 

Video Script:

Make Max SPL Measurements to the AES 75 Standard using M-Noise

Our pre-written test sequence automates measurement to the new AES 75 standard for Maximum SPL, removing subjectivity, increasing reliability, and saving time.

This standard details a method for measuring maximum linear sound levels of a loudspeaker system or driver using M-Noise, a test signal specifically developed to emulate the dynamic characteristics of music. Clearly defined limits for linear frequency response and coherence determine the Max SPL level and remove any measurement ambiguity.

Implementing the standard manually relies on an operator’s subjective judgment of real time spectrum analyzer data. It also requires multiple iterations of a measurement, therefore is labor intensive.

Our test sequence is fully automated. We use the same test signal and calculations outlined in the standard, but automated analysis steps objectively calculate the measurements and drive the next steps in the procedure.

Let’s take a look.

Here, you can see we are using the freely available M-Noise test signal introduced by Meyer Sound. This stimulus features a relatively constant peak level as a function of frequency, but a diminishing RMS level with increasing frequency.

First, we use a test signal approximately 20dB below our expected Max SPL to obtain a provisional linear frequency response, linearity, coherence and signal to noise ratio. We then increase the test level by 3dB, and compare the results to the initial value. The results must be within +/- 1dB, have a coherence of at least 97% and a signal to noise ratio 15dB or higher so that we know we are operating in the speaker’s linear region and our signal to noise ratio is sufficient for accurate measurements.

Next we automatically increase the test level by 3dB and compare it to the initial results, normalized to the current test level. Multiple measurement iterations take place until one of the ‘stop’ conditions is reached. These conditions are either:

  • the live measurement differs from the linear frequency response by at least 2 dB over at least two octaves
  • the live measurement differs from the linear frequency response by at least 3 dB anywhere, or
  • the Coherence Reduction Target is met –  this means the signal to noise ratio is 10dB or less and/or the coherence is 91% or less.

When one of these limits is reached, the sequence then reduces the test level to the last level that passed and repeats the measurements in 1dB increments to find the precise Level at which the response deviates from the base level.

Once this level is established, the device enters a burn-in process, where the M Noise stimulus is played through the DUT for five minutes and fifty-three seconds and again compared to the initial result. This long duration measurement is also used to generate peak, rms and A weighted RMS sound levels. If the response curve remains consistent, this curve is the Max SPL curve according to the standard. If it is not within the acceptable limits, the device is cooled down and the tests repeated with a longer stimulus duration.

All the operator needs to do is enter any stimulus limits based on the operating range of the DUT into the sequence before starting, then return when the test is complete.

As well as saving time, SoundCheck mathematically calculates the data in analysis steps within the sequence, which avoids the subjectivity of relying on operator interpretation of real-time spectrum analyzer outputs. This increases repeatability and confidence in the results. This sequence is available free of charge on our website. Check it out!

25: Confidence Limits (working from home with SoundCheck)

Following questions from some SoundCheck users who had watched #3 in this series (sequence optimization), Steve Tatarunis takes a look at the measurement confidence function in SoundCheck, and explains the trade-offs between speed and accuracy when choosing a step size to optimize your measurement.

24: Instrument Highlights (working from home with SoundCheck)

Anastassia Tolpygo demonstrates some neat features of virtual instruments that you may not have seen before. These enable you to make a quick distortion measurements, accurately measure frequencies at very high resolution, and plot and save curves over time using just the virtual instruments without the need for a sequence step.

23: Offline Statistics (working from home with SoundCheck)

In this short video, Steve Tatarunis takes a deeper dive into the Offline Statistics functionality in SoundCheck, demonstrating how you can run statistical analyses from home on data from your production facility or overseas contract manufacturer. No hardware required – just your computer with SoundCheck and some measured data!

22: Save To Office (working from home with SoundCheck)

What better way to perfect your data output options than when you have some uninterrupted SoundCheck time at home! SoundCheck saves data in many different user defined ways, and saving to Office apps such as Word and Excel is just one of them. Most people just use the default settings to create a quick visual output, but the key to making your data work for you is Custom Templates and Autosave Steps. Watch this short video to learn more.

21: Limits (working from home with SoundCheck)

While you are working from home, optimize your SoundCheck sequences by tweaking your limits for optimum quality and yield. Devin Vaillancourt explains the different types of limits available (absolute, floating, aligned, and more), and once you have explored the different types, you can test out different limit configurations using saved data from products that you have already measured.

20: SoundMap (working from home with SoundCheck)

Cam Ruffle-Deignan walks you through a demonstration of SoundMap, SoundCheck’s Time-Frequency analysis module, showing how Short Time Fourier Transform, CSD, Wigner Ville and Wavelet analyses can be used for detailed offline analysis of devices. Time-Frequency analysis methods are useful for impulse response analysis and detection of loose particles and Rub & Buzz in loudspeakers and identification of transient effects such as drop out in digital devices including VoIP and Bluetooth headsets.

19: Exploring Custom Steps (working from home with SoundCheck)

Learn how to level-up your SoundCheck sequence using Custom Steps – LabVIEW VIs that can be used as steps in SoundCheck sequences to implement operations that cannot be done natively in SoundCheck. A selection of custom step templates built into SoundCheck accelerates the development of custom steps with examples that allow you to open and close virtual instruments, control turntables, read/write serial numbers and even open any .exe file such as a Python script. Devin Vaillancourt demonstrates how to use this powerful feature.

18: Displaying multiple data sets on one graph (working from home with SoundCheck)

There are many times when it is desirable to display multiple data sets on one graph – for example, testing multichannel devices, comparing devices, comparing measurements to the fundamental, comparing EQ curves, etc. In this short video, Steve Tatarunis demonstrates how to display multiple curves on one graph, both by simply dragging and dropping and also automatically via a sequence.