A New Method for Transient Distortion Detection

Transient distortion, or ‘loose particle’ measurement, is an important loudspeaker production line quality control metric that identifies and facilitates troubleshooting of manufacturing issues.

This paper introduces a new enhanced loose particle measurement technique that discriminates more accurately and reliably than current methods. This new method introduces ‘prominence’ after envelope detection, a new metric for audio measurements, that effectively isolates transient distortion in the presence of periodic distortion. This technique also offers the unique ability to listen to the isolated transient distortion waveform which makes it easier to set limits based on audibility and has widespread applications.

Authors: Steve Temme, Rahul Shakya and Jayant Datta, Listen, Inc.
Presented at 155th AES Conference (October 2023) New York, NY

Full Paper

 

Paper Introduction

Transient distortion, or ‘loose particle’ measurement, is a valuable quality control metric because it identifies non-periodic distortion, for example, rattling parts, separately from periodic distortion such as rubbing or buzzing parts. This facilitates troubleshooting of manufacturing issues. This paper introduces a new transient distortion measurement technique that is more accurate and reliable than current methods. In addition to improved performance, this new algorithm also aids understanding of the correlation between measurement results and audibility, since it is possible to isolate and listen to just the transient distortion artifacts. Although this analysis method was developed for measuring loose particles in loudspeaker drivers, it is also valuable for measuring rattling parts such as buttons, fasteners, and loose wires on various audio devices, and measuring impulsive distortion or Buzz, Squeak and Rattle (BSR) in automotive audio applications [1].

What is Transient Distortion? Why does it matter?

Transient distortion is caused by random clicking, popping, and other noises in the time domain. In a speaker or headphone driver, this might be caused by foreign particles such as glue or magnet fragments trapped in the gap behind the diaphragm or dust cap. In a device such as a smart speaker, transient distortion might come from a loose volume control button on the device that rattles when sound is played. In an automotive application, it could be characterized as buzz, squeak and rattle from loose wires, screws or fasteners in a car door that the loudspeaker is mounted in. In all cases, the sound is undesirable, so devices that exhibit such faults should be identified and rejected.

In the recorded time waveform, transient distortion faults appear as impulsive noises added on the stimulus wave. These impulses are not related to the frequency of the stimulus, but rather to the vibration caused by the displacement amplitude of the diaphragm. The transient distortion is more frequent and significant when the speaker is driven near or below its resonant frequency, where the displacement of the diaphragm is the greatest.

Although the sound – a random clicking, buzzing or popping noise – can sometimes sound similar to higher order harmonic distortion (Rub & Buzz), such defects are not clearly reflected in the frequency spectrum of the waveform. Figure 1 shows a waveform with transient distortion, and the corresponding frequency spectrum. The vertical black line represents the stimulus frequency and the orange broadband noise spectrum indicates the transient distortion. Transient distortion is best identified at the time the transients occur, unlike Rub & Buzz distortion which is best identified by the frequency at which it occurs [3].

The entire paper also covers:

Prior Measurement Methods

The new algorithm – comparison and results

Conclusions

Download paper

In addition to this paper, please also check out the Enhanced Loose Particles Webpage and our detailed video explanation of how this algorithm works.

Enhanced Perceptual Rub & Buzz Measurement for Testing Automotive Loudspeakers

Loudspeaker Rub & Buzz faults are a problem for automotive manufacturers as they sound harsh and immediately give the perception of poor quality. There are two places such faults can occur – during speaker manufacturing and installation of the speaker in the car. A buzzing loudspeaker in a car is disappointing to a customer and is costly to replace. It is also challenging for a service center to determine exactly where the buzzing is coming from and whether it is caused by a faulty loudspeaker or bad installation. Perceptual distortion measurements are often considered the holy grail of end-of-line testing because rejecting speakers with only audible faults increases yield. Although such measurements have been around since 2011, production line adoption has been slow because until now, sensitivity to background noise has made limit-setting challenging. In this paper, a new algorithm is introduced that uses advanced technology to reduce the impact of background noise on the measurement and offer more repeatable results. This facilitates limit setting on the production line and makes it a truly viable production line metric for increasing yield. This same metric may also be used for end-of-line automotive quality control tests. Results from various algorithms will be shown, and their correlation to subjective and other non-perceptual distortion metrics explained.

Author: Steve Temme, Listen, Inc.
Presented at 2022 AES Automotive Conference, Dearborn, MI

Full Paper

 

Introduction

The automotive industry’s stringent quality expectations make end-of-line quality testing on automotive speakers and drivers absolutely critical. End-of-line tests typically measure a range of parameters including frequency response, THD, and polarity. Manufacturing-introduced defects such as Rub & Buzz and Loose Particles are also measured. Reliable, automated testing has been available for decades now, and most large manufacturers rely on these software-based systems for identification and rejection of defective products. While these tests do an excellent job of identifying defective units, there is always a certain level of false rejection where units with some distortion fail even though it is completely inaudible to the human ear. From a manufacturing perspective, higher yields and therefore greater profitability is always desirable.

Perceptual Distortion Measurements

This has driven the development of perceptual distortion measurements – automated measurements that replicate the human hearing to detect only audible distortion defects. Such metrics increase production line yield by passing products with inaudible distortion, as the product will still sound exactly as the manufacturer intended. Perceptual methods are very simple to configure for production line use. Since they return a result in Phons, an absolute measurement that can be easily correlated to the listener’s threshold of hearing, the operator can set a fixed limit across the board, regardless of product. Naturally, the price point and quality expectations for the product may influence the level of distortion that is deemed acceptable.

Perceptual Distortion Algorithms

Our algorithm, introduced in 2011, was the first commercial perceptual distortion metric, although in the past couple of years, other test system manufacturers have also started to offer perceptual distortion tests. It offers excellent correlation with human hearing and performs well in laboratory tests. However, like the human ear, repeatability decreases in the presence of background noise. This is not a failure of the algorithm as such, but an indication that the algorithm performs just like a human listener; when background noise is high, audible distortion is masked. This limitation restricts the value of such algorithms on the production line, as with today’s high-volume manufacturing, there is only time for one fast test sweep. If this sweep gets a different result under changing background noise conditions, limit setting becomes challenging, and repeatability and reliability is decreased. Similar algorithms from other test system manufacturers also suffer from the same problems.

New Perceptual Distortion Algorithm Development

This paper details efforts to create an algorithm that hears like a human in quiet conditions, e.g. in a living room or passenger automotive cabin, under the less-than-perfect conditions of a manufacturing environment where considerable and varying background noise may be present. In other words, a perceptual model that is more independent and reliable than the human ear when it comes to noisy environments. The resulting new algorithm overcomes these limitations to offer repeatable end-of-line test results, even in noisy environments. It incorporates noise reduction techniques and enhanced perceptual filters to overcome the reliability and high frequency masking issues of earlier versions. In short, the algorithm offers the performance of an ‘enhanced’ human ear – it detects distortion like an ear in a quiet environment, even when there is background noise. This makes it a viable solution for production line use.

In this paper we explain how the algorithm works, demonstrate how the results compare with earlier perceptual algorithms and show its correlation with human hearing and conventional distortion algorithms. We also compare its performance in the presence of background noise to other perceptual algorithms by adding recorded factory background noise to the signal before passing it through the algorithms.

Read More

More about Listen’s enhanced Perceptual Rub & Buzz algorithm

More about in-car measurement of  impulsive distortion / Buzz, Squeak and Rattle.

A New THD+N Algorithm for Measuring Today’s High Resolution Audio Systems

In this paper, a mathematical definition of Total Harmonic Distortion + Noise suitable for testing high-resolution digital audio systems is presented. This formal definition of the “distortion analyzer” mentioned in AES17 defines THD+N as the RMS error of fitting a sinusoid to a noisy and distorted sequence of measurements. We present the key theoretical result that under realistic conditions a modern THD+N analyzer is well-described by a Normal probability distribution with a simple relationship between relative error and analysis dwell time. These findings are illustrated by comparing the output of a commercial distortion analyzer to our proposed method using Monte Carlo simulations of noisy signal channels. We will demonstrate that the bias of a well-designed distortion analyzer is negligible.

Authors: Alfred B. Roney The Mathworks, Inc. (formerly Listen, Inc.) Steve Temme, Listen, Inc.
Presented at AES 2018, New York, NY.

Full Paper

Advances in Impedance Measurement of Loudspeakers and Headphones

Impedance measurement is often the sole electrical measurement in a battery of QC tests on loudspeakers and headphones. Two test methods are commonly used, single channel and dual channel. Dual Channel measurement offers greater accuracy as both the voltage across the speaker (or headphone) and the reference resistor are measured to calculate the impedance. Single Channel measurement methods are more commonly used on the production line because they only require one channel of a stereo soundcard, which leaves the other free for simultaneous acoustic
tests. They are less accurate, however, due to the test methods making assumptions of constant voltage or constant current. In this paper we discuss a novel electrical circuit that offers similar impedance measurement accuracy compared to complex dual channel measurement methods but using just one channel. This is expected to become popular for high throughput production line measurements where only one channel is available as the second channel of the typical soundcard is being used for simultaneous acoustic tests.

Authors: Steve Temme and Tony Scott
Presented at the 135th AES Conference, New York 2013

Full Paper

Measurement of Harmonic Distortion Audibility Using A Simplified Psychoacoustic Model – Updated

A perceptual method is proposed for measuring harmonic distortion audibility. This method is similar to the CLEAR (Cepstral Loudness Enhanced Algorithm for Rub & buzz) algorithm previously proposed by the authors as a means of detecting audible Rub & Buzz which is an extreme type of distortion[1,2]. Both methods are based on the Perceptual Evaluation of Audio Quality (PEAQ) standard[3]. In the present work, in order to estimate the audibility of regular harmonic distortion, additional psychoacoustic variables are added to the CLEAR algorithm. These variables are then combined using an artificial neural network approach to derive a metric that is indicative of the overall audible harmonic distortion. Experimental results on headphones are presented to justify the accuracy of the model.

Authors: Steve Temme, Pascal Brunet and Parastoo Qarabaqi
Presented at the 51st AES Conference, Helsinki, Finland, 2013

Full Paper

Measurement of Harmonic Distortion Audibility Using A Simplified Psychoacoustic Model

A perceptual method is proposed for measuring harmonic distortion audibility. This method is similar to the CLEAR (Cepstral Loudness Enhanced Algorithm for Rub & buzz) algorithm previously proposed by the authors as a means of detecting audible Rub & Buzz which is an extreme type of distortion[1,2]. Both methods are based on the Perceptual Evaluation of Audio Quality (PEAQ) standard[3]. In the present work, in order to estimate the audibility of regular harmonic distortion, additional psychoacoustic variables are added to the CLEAR algorithm. These variables are then combined using an artificial neural network approach to derive a metric that is indicative of the overall audible harmonic distortion. Experimental results on headphones are presented to justify the accuracy of the model.

Authors: Steve Temme, Pascal Brunet and Parastoo Qarabaqi
Presented at the 133rd AES Convention, San Francisco, 2012

Full Paper

Practical Implementation of Perceptual Rub & Buzz Distortion and Experimental Results

In a previous paper [1], we demonstrated how an auditory perceptual model based on an ITU standard can be used to detect audible Rub & Buzz in loudspeakers using a single tone stimulus. In this paper, we discuss a practical implementation using a stepped sine sweep stimulus and present detailed experimental results on loudspeakers including comparison to human listeners and other perceptual methods.

Authors: Steve Temme, Pascal Brunet and Brian Fallon
Presented at the 129th AES Convention, San Francisco, 2010

Full Paper

Practical Measurement of Loudspeaker Distortion Using a Simplified Auditory Perceptual Model

Manufacturing defects in loudspeaker production can often be identified by an increase in Rub & Buzz distortion. This type of distortion is quite noticeable because it contributes an edgy sound to the reproduction and is annoying because it often sounds separate or disembodied from the fundamental signal. The annoyance of Rub & Buzz distortion is tied intimately to human perception of sound and psychoacoustics. To properly implement automated production-line testing of loudspeaker Rub & Buzz defects, one has to model or imitate the hearing process using a sufficiently accurate perceptual model. This paper describes the results of a Rub & Buzz detection system using a simplified perceptual model based on human masking thresholds that yields excellent results.

Authors: Steve Temme, Pascal Brunet and D.B. (Don) Keele
Presented at the 127th AES Convention, New York, 2009

Full Paper

Evolution of Time-Frequency Analysis Methods and their Practical Applications

Time-Frequency analysis has been in use for more than 20 years and many different Time-Frequency distributions have been developed. Four in particular, Short Time Fourier Transform, Cumulative Spectral Decay, Wavelet and Wigner-Ville have gained popularity and firmly established themselves as useful measurement tools. This paper compares these four popular transforms, explains their trade-offs and discusses how to apply them to analyzing audio devices. Practical examples of loudspeaker impulse responses, loose particles, and Rub & Buzz defects are given as well as a demonstration of their application to common problems with digital/analog audio devices such as Bluetooth headsets, MP3 players and VoIP telephones.

Authors: Pascal Brunet, Zachary Rimkunas and Steve Temme
Presented at the 123rd AES Convention, New York, 2007

Full Paper

A New Method for Measuring Distortion using a Multitone Stimulus and Non-Coherence

A new approach for measuring distortion based on dual-channel analysis of non-coherence between a stimulus and response is presented. This method is easy to implement, provides a continuous distortion curve against frequency, and can be used with a multitone stimulus, noise, or even music. Multitone is a desirable test signal for fast frequency response measurements and also for assessing system nonlinearities. However, conventional single-channel multitone measurements are challenging because the number of intermodulation tones grows rapidly with the number of stimulus tones and makes it extremely difficult to separate harmonics from intermodulation products. By using dual-channel measurement techniques, only well-known, standard signal processing techniques are used, resulting in simplicity, accuracy and repeatability.

Authors: Steve Temme and Pascal Brunet
Presented at the 121st AES Convention, San Francisco, 2006

Full Paper