Transient multi-tone test signal and method for audio speakers

Abstract

A transient multi-tone test signal is generated to detect piano noise distortion of speaker playback. The test signal has a number of formant tones, such as three, similar to a piano chord and are selectable such that the frequencies and harmonics are distributed in frequency. The test signal provides a basis to define normal speaker operating performance and allows a user to define speaker response that is outside of the normal bounds by highlighting the piano noise distortion detected during speaker playback of the test signal.

Description

[0001] Cross-reference to related applications

[0002] This application claims priority to U.S. Provisional Patent Application No. 63 / 030,812, filed May 27, 2020, and U.S. Provisional Patent Application No. 63 / 185,567, filed May 7, 2021, which are incorporated herein by reference. Technical Field

[0003] The embodiments generally involve generating multi-tone test signals for testing audio loudspeakers to detect piano noise distortion. Background Technology

[0004] The demand for high-quality audio playback in small portable devices such as mobile phones, tablets, and laptops has led to higher quality control standards for small (micro-transducer) loudspeakers. Due to their small size, tight tolerances, and limited air volume, even tiny manufacturing defects in such loudspeakers can result in significant audio degradation.

[0005] Piano noise is a type of distortion in loudspeakers that has been recognized as a common problem primarily affecting miniature transducers used in small, portable devices. Piano noise can be a very noticeable distortion, appearing even at relatively low output levels when reproducing certain types of audio signals. As the name suggests, this type of distortion primarily affects the reproduction of piano tones, but other types of sounds can also appear to have this distortion, such as chimes (e.g., the well-known tone played when a computer starts up) or content with a similar timbre.

[0006] Smaller speakers are particularly prone to this distortion problem because the volume of enclosed air extending behind the transducer in a speaker module, and the air around the speaker in portable devices, is limited. This problem is even more pronounced in miniature transducers because the ratio of the driver's radiating surface area to the air volume within the housing is much higher than in more conventional speaker systems, resulting in highly compressed trapped air. If air is drawn in and expelled from the housing through small openings, acoustic artifacts (such as those indicated by a hum) can easily form. Because these artifacts tend to be frequency-separated from the audio or content signal, their audibility is enhanced, leading to perceptible distortion.

[0007] Current methods for testing manufactured audio loudspeakers typically involve using simple test tones to provide a basic pass / fail test for the loudspeaker's functionality. These methods offer no mechanism for identifying subtle or minute distortions that might not cause the device to malfunction, but could still render the loudspeaker useless in most playback applications. Summary of the Invention

[0008] The embodiment includes a method for testing piano noise distortion of an audio loudspeaker by generating transient test signals with different formant tones, the formant tones being configured to allow frequencies and harmonics to remain distributed across frequencies, wherein the test signal occupies a stimulus region within the audio spectrum. The test signal is applied to the loudspeaker, and the loudspeaker's microphone response is measured to capture the frequency response within an analysis region of the audio spectrum separate from the stimulus region. The measured microphone response is compared to a derived baseline value to determine whether the loudspeaker's performance is acceptable or unacceptable, for reporting to the user.

[0009] The embodiments also relate to a method for testing loudspeakers, including obtaining standard frequency response measurements and piano noise measurements of a sample number of loudspeakers within a batch, deriving a threshold for distinguishing between qualified and unqualified loudspeakers regarding piano noise distortion, and generating a transient test signal having multiple selectable formant tones similar to piano chords, such that frequencies and harmonics remain distributed across frequencies. The test signal is shaped into a pulse signal with a burst power applied within a shaped decay period, which is then applied to each loudspeaker. The response of each loudspeaker is then recorded and used to derive the piano noise distortion level of each loudspeaker. The derived distortion level is compared to the threshold to determine whether each loudspeaker performs well or poorly, and the results can be reported to a user.

[0010] The embodiments also relate to a system that generates a test signal having selectable multiple different formant tones similar to piano chords, such that frequencies and harmonics are maintained in a frequency distribution, wherein the test signal occupies a stimulating region within the audio spectrum, and is applied to a loudspeaker. The system obtains baseline measurements of standard audio response and piano noise distortion. The test signal applied to the loudspeaker under test can be a pulse signal shaped with respect to power, decay time, and decay envelope shape to provide a basis for identifying piano noise components within the loudspeaker's frequency response. The loudspeaker's output is provided to an analyzer that captures the loudspeaker's frequency response and compares the measured microphone response with the baseline measurements to determine whether the loudspeaker's performance is acceptable or unacceptable, which is then relayed to the user.

[0011] Therefore, such embodiments use short-term (transient) test signals with a spectrum between separate stimulus and analysis bands to clearly show the test signal separated from any resulting distortion artifacts. The frequency intervals and levels of the various formant tones are configured to excite harmonic and non-harmonic related tones, which, if caused by a defective loudspeaker, can be clearly shown in the analysis area. Attached Figure Description

[0012] In the following figures, the same reference numerals are used to refer to the same elements. Although the following figures depict various examples, one or more implementations are not limited to the examples depicted in the figures.

[0013] Figure 1 A defective small-scale transducer loudspeaker, which can be tested using a three-tone test signal according to some embodiments, is shown.

[0014] Figure 2 The time response of a transient multi-tone test signal according to some embodiments is shown.

[0015] Figure 3 According to some embodiments Figure 2 The spectrum of the multi-tone test signal.

[0016] Figure 4 It is shown according to some embodiments Figure 3 Table 400 shows the different values ​​of the three resonance peaks of the test signal.

[0017] Figure 5 This is an illustration based on some embodiments. Figure 4 The table shows the curves of the resonance peak distribution.

[0018] Figure 6A Based on the use of some embodiments Figure 3 Example response diagram of a small loudspeaker tested with a test signal.

[0019] Figure 6B The diagram shows only the response graph of a good speaker. Figure 6A The curve graph.

[0020] Figure 6C The diagram shown only illustrates the response of the defective speaker. Figure 6A The curve graph.

[0021] Figure 7 It is a graph showing the effect of windowing the amplitude of the test signal relative to time according to some embodiments.

[0022] Figure 8 A method for testing a loudspeaker using a multi-tone test signal, according to some embodiments, is shown.

[0023] Figure 9 A test system is shown that implements a test method using a multi-tone test signal according to some embodiments. Detailed Implementation

[0024] The embodiments relate to generating and using test signals to detect and identify audio distortions manifested as piano noise, particularly for microtransducer loudspeaker systems. Such test signals can be used in a holistic test methodology to evaluate loudspeakers commercially produced by loudspeaker manufacturers. Any of the described embodiments can be used alone or in any combination with each other. While the various embodiments may have been inspired by various shortcomings of current and known solutions, which may be discussed in the specification, these embodiments do not necessarily address any of these shortcomings. Different embodiments may address different shortcomings, and some shortcomings may only be partially addressed.

[0025] For the purposes of this specification, the following terms have the associated meanings: The term "loudspeaker" or "amplifier" refers to an audio playback loudspeaker having a cabinet or enclosure housed within one or more drivers, wherein the term "driver" refers to a separate audio transducer that converts an electro-audio signal into sound waves, and can be implemented as a cone, dome, compression driver, miniature loudspeaker, or planar driver, and can be a full-range driver or a driver configured to reproduce a specific frequency range, such as a tweeter, mid-range driver, woofer, subwoofer, etc. The term "cabinet" refers to a loudspeaker enclosure or housing that houses one or more transducers (or drivers), and is typically completely enclosed to acoustically isolate the rear of the transducers, but can also be ventilated or partially open if certain audio response characteristics are required. The enclosure may include part or all of a device housing (e.g., a mobile phone or portable computer) that integrates the loudspeaker within its housing. The term "artifact" refers to an anomaly caused by distortion, such as in an audio signal, and may manifest as unwanted but perceptible audio components at frequencies or within a frequency range in the loudspeaker response.

[0026] As mentioned above, small loudspeakers can be affected by manufacturing or material defects that can cause micro-leakage between the enclosed air volume behind the transducer and the air surrounding the speaker. This problem is often more significant for miniature transducer loudspeakers because the ratio of the transducer's radiating surface to the air volume within the housing is large, resulting in high compression of the trapped air. If severe enough, this leakage can lead to auditory distortion due to audio artifacts that separate the audio spectrum from the content itself.

[0027] Figure 1 A defective small-scale transducer loudspeaker, which can be tested using audio signals and testing methods according to some embodiments, is illustrated. Figure 1As shown, the loudspeaker 100 includes a transducer (or diaphragm) 102 fixed to a frame 104. The transducer comprises a thin film made of paper, laminated material, or plastic (e.g., polypropylene), which is suspended at its edges by a flexible structure (suspension stage) 106. If there are any holes or cracks in the sensor material, the loudspeaker will not function properly. Proper integration or bonding of the diaphragm material 102 and the suspension stage 106 to each other is also crucial for proper performance. They must also be aligned and correctly bonded to the frame 104. Any gaps between the diaphragm, the frame, and the frame can cause the loudspeaker to fail. Like holes in the diaphragm, such air leakage typically manifests as an audible clicking or similar annoying sound during audio playback.

[0028] Figure 1 An exemplary defect in speaker 100 is shown, in the form of an air gap 108 in one corner of the speaker, caused by an incomplete fit between the mounting platform and the frame. Depending on the size and configuration of the speaker, even the smallest gap can cause auditory distortion. The size, composition, shape, and configuration of speaker 100 are examples of small speakers used in portable device products such as mobile phones. Such speakers are typically shaped to fit confined areas and can therefore be elongated (as shown) to provide maximum transducer area for small, narrow spaces within the device. For illustrative purposes, Figure 1 The Speaker 100 is a small speaker with a transducer on the order of 13mm x 38mm, housed in a small casing of approximately 4 cubic centimeters. Speakers of this size are typically used in 12-inch to 15-inch laptops, but even smaller speakers are commonly found in mobile phones and tablets.

[0029] The characteristic of a small loudspeaker is that the area of ​​the transducer 102 is relatively large compared to the area and volume around the loudspeaker 100, so even minor manufacturing defects can cause a certain degree of auditory distortion.

[0030] It should be noted that although the embodiments are illustrated and described with respect to small or miniature loudspeakers, the embodiments are not limited thereto, and loudspeakers of any size or configuration can be tested using the test signals and methods described herein.

[0031] In the mass production of small loudspeakers, due to cost constraints and the difficulty of conducting extensive visual inspections, many loudspeakers carry defects such as… Figure 1 The defects shown are very common when the product leaves the factory. For example, Figure 1The loudspeaker 100 shown exhibits a common problem of air leakage caused by one or more small gaps in the frame where the lead wires are routed. Since miniature loudspeakers often lack wiring to the rear volume (due to limited speaker depth), the wires must frequently protrude from the front of the loudspeaker and from the underside or edge of the transducer 102. Other types of defects are also common in small loudspeakers, such as defective or loosely fitted mountings, pinholes in the transducer, etc. Any such defect typically creates an area where a tiny leak between the trapped air behind the diaphragm and the outside air causes problems. Piano noise distortion is often associated with this air leakage problem and can be distinguished by other sources of higher harmonic distortion, such as mechanically induced vibrations (e.g., friction and hum).

[0032] In some cases, loudspeakers are designed with vents or ports to house components (e.g., motors) and / or to shape the sound of the loudspeaker. In such cases, blocking any intended vents can cause the loudspeaker to produce the same or similar types of distortion.

[0033] The embodiments relate to a testing method that uses a unique test signal to effectively detect air leakage defects in a loudspeaker (e.g., loudspeaker 100). The test signal is configured to detect piano noise distortion, which is typically caused by minute air leaks from the rear of the transducer to the surrounding area. Therefore, the embodiments include a test signal developed to diagnose piano noise problems associated with air leakage in small transducers.

[0034] Test signal derivation

[0035] Piano tones are typically characterized by high onset and rapid decay, with each frequency (note) properly defined, whether played solo or in a chord. Revealing piano noise problems in loudspeakers requires exciting it with an appropriate type of signal. It has been found that audio content containing piano chords highlights piano noise problems in several loudspeakers found to be defective. Based on this finding, a test signal for reproducing piano chords has been developed. This test signal takes into account the formant structure and time envelope of the signal, but omits higher harmonics, thus producing a useful test signal. Higher harmonic content often gives piano sounds special characteristics. However, to analyze artifacts caused by potential defects, the test signal deliberately omits higher harmonic components so that the upper spectrum is empty, and these artifacts are known to occupy these frequency bands.

[0036] The derivation of the test signal allows users to excite the loudspeaker and collect distortion data, and also clearly hear problems, such as in the case of reproducing piano content, thus enabling the establishment of some assessments of acceptable distortion thresholds. Therefore, a representative test signal highlights distortion and simplifies the evaluation and quantification process for testing large samples or high-volume production loudspeakers.

[0037] In one embodiment, certain observations and evaluations obtained during the experiment are used to derive an audio test signal. This process is used to develop a piano noise test signal containing multiple (e.g., three) different formant frequencies. Piano notes typically consist of a formant (dominant) tone followed by a series of odd and even harmonics. While individual notes can excite noise, chords (multiple formants) make the distortion more noticeably perceptible. Furthermore, it has been found that higher harmonics, which typically give the piano different sounds (Steinway vs. Bosendorfer), are of less value in exciting distortion. In deriving the test signal, numerous piano samples were analyzed, and the temporal and harmonic structures were evaluated. It has generally been found that artifacts produced by piano noise distortion are harmonic-related, but primarily excite higher orders (e.g., 7th and higher). It has been discovered that the properties of air itself, particularly its bulk modulus, can change due to high compression, resulting in secondary distortion that excites lower harmonics and intermodulation (sum and difference) tones.

[0038] As mentioned above, the derived piano noise test signal is a transient irregular audio signal containing three formant frequencies. Figure 2 The time response of an audio test signal is shown in some embodiments. Figure 2 Graph 200 shows the change in amplitude of test signal 202 over time. To create short-term events similar to piano notes, a linear decay window was used to gradually decrease the amplitude of the test signal from 1V to 0V over 0.5 seconds. Figure 2 An example of a test signal pulse is shown, and many other test signal pulses can be used. The relevant parameters of this pulse are pulse power, decay time, and decay envelope (e.g., cone-down, cone-up, square, pulse, etc.). Therefore, Figure 2 The example shows a test signal pulse containing 0.5 seconds with a linear decay cone, but other envelope decay times and gradient profiles can also be used.

[0039] Test signal 202 is provided as a short burst signal to ensure low power consumption of the speaker under test. The test signal is configured to mechanically compress the speaker and force air through any air gaps or holes that may be present. The burst signal contains high-amplitude formant tones to highlight piano noise issues. The test signal is provided as a burst signal rather than a continuous signal to reduce the risk of any thermal changes or malfunctions in the speaker.

[0040] To make the test signal applicable to loudspeakers of different frequency ranges, a means of selecting an appropriate range for the test tone pulses was established. A "tritone" (three formant) signal was generated such that the frequency of the first tone was at a standard 1 / 3 octave band, and subsequent tones were set to multiples based on the golden ratio method. The frequencies of the formants could be arbitrarily chosen based on chords observed in shared excerpts of piano music, or by selecting harmonically unrelated formant tones that could spread more widely in the spectrum. Although an embodiment was described for a tritone test signal, other numbers of formant tones are also possible.

[0041] Regarding the time-domain elements of the test signal, such as Figure 2 As shown, the test signal 202 is a non-stationary short-term or transient signal, also known as a signal "burst". During the test, the test signal can be applied to one or more speakers as a single burst or a limited series of bursts (e.g., 8 to 24 bursts).

[0042] Figure 3 In some embodiments Figure 2 The frequency domain representation of the test signal. Figure 3 Graph 300 shows the frequency response of the test signal 306 in dB. The test signal 306 is generated within a selected frequency range (e.g., 100 Hz to 2 kHz) designated as the “stimulation region” 302 and applied to the loudspeaker under test. The loudspeaker’s response is then observed in the “analysis region” 304 (e.g., 2 kHz to 10 kHz). As described above, the analysis region 304 is created by omitting the higher harmonics of the test signal.

[0043] like Figure 3 As shown, the test signal 306 was formed to include three distinct resonant peaks (shown as signal peaks) 308. Using the golden ratio, a resonant peak ratio of 1:1.09:1.17 was used to give a regular logarithmic frequency interval, which will be reflected in the distribution of harmonics. It has been shown that this ratio is sufficient to excite any suitable loudspeaker when applied within the stimulation region 302. Figure 4 In some embodiments, it is shown Figure 3Table 400 shows the different values ​​of the three formant peaks 308. Table 400 shows the formant tones generated based on the frequency of the first tone using the described ratio at a standard 1 / 3 octave band level. This is scalable in bandwidth, so for micro-transducers, the system can be configured to start stimulation generation, for example, at 200 Hz, to avoid exciting the loudspeaker outside the range in which it will be used.

[0044] Although an embodiment has been described with regard to a test signal having three formant tones, the embodiment is not limited thereto, and other practical numbers of formant tones can be used, such as from 2 to 24 formant tones. It has been found that three formant tones are sufficient to produce satisfactory discrimination of the worst-case excitation band for a given loudspeaker resonance relative to the internal air pressure of a typical microspeaker; however, depending on the loudspeaker design and defect characteristics, fewer or more formant tones may be used. In one embodiment, the number of formant tones and their relative spacing can be user-selectable to allow further adjustment of the test signal to suit a particular type of loudspeaker.

[0045] Figure 5 As shown in some embodiments, Figure 4 Table 400 shows the curves illustrating the resonant peak distribution. Curve 500 shows that, on a logarithmic frequency scale, the resonant peaks remain distributed across the frequencies. This process excites harmonic and non-harmonic related tones and avoids overlapping stimulation between steps. Therefore, the analysis region 304 should include harmonics and intermodulation artifacts that would not be excited using a single tone.

[0046] In the example embodiment, Figure 3 The test signals were applied to laptop computer modules containing small drives (e.g., drives measuring 13mm × 38mm within a 4 cubic centimeter casing). The module was intentionally damaged by drilling holes in the front and rear of the casing to simulate real-world defects. The frequency of the test signals was increased in 1 / 3 octave increments to study the frequency range that produces the maximum artifact level in the analysis region between 2kHz and 10kHz. Figure 6A This is an example response curve graph 600 for this test setup. Figure 6A As shown, graph 600 illustrates a test signal 602 applied within the stimulation region (200 Hz to 2 kHz), and the resulting example response curves within the analysis region (2 kHz to 10 kHz). Curve 608 of graph 600 shows an example response of a completely sealed loudspeaker (i.e., without defects), while curve 610 shows an example response of a loudspeaker with a hole at the rear. From Figure 6A As can be seen, the response curve 610 of the defective loudspeaker shows a significant signal in the analysis region of the spectrum, which is on the order of +20 to +30 dB higher than that of the defect-free loudspeaker 608.

[0047] Figure 6B It shows Figure 6A The graph shows the response of a good loudspeaker. From Figure 6A As can be seen, the test signal 620 is confined to the stimulation region up to 2kHz, while the upper frequency region is empty except for any minute or residual noise. Figure 6B As shown, when the test signal is applied to a good loudspeaker, the analysis area should not show any obvious artifacts. Figure 6C It shows Figure 6A The graph shows the response of the defective loudspeaker. In this case, the test signal caused noticeable artifacts 640 in the analysis region above 2 kHz.

[0048] It should be noted that curve 610 is an example curve for a defective loudspeaker with air leakage caused by gaps or holes in the diaphragm, surround, or the interface between them. This defect can be severe enough to cause very noticeable artifacts in the analysis area. The figure is shown for illustrative purposes, and the embodiment can be applied to other defects, such as any manufacturing deviations that cause air leakage in the loudspeaker, which may produce different or smaller artifact patterns.

[0049] refer to Figure 6A In addition to the spectrum of the loudspeaker module under test, the transducer displacement curves (in dBr) were superimposed to investigate any relationship between distortion and displacement and internal housing pressure. Curve 606 shows the diaphragm displacement corresponding to the response curve and is shown as a reference relative to the lower curve 602, which shows the acoustic response of a defective device generating noise in the upper frequency range. Figure 6A As can be seen, high distortion is usually related to the excitation frequency, which is centered at the point where the driver displacement is the greatest and therefore the housing pressure is the greatest. The test tone with the first resonance peak at 400 Hz is a revealing test signal for this particular loudspeaker module.

[0050] Regarding the frequency band affected by piano noise, some experimental data has shown that for loudspeakers with large driver areas and small back volumes, two distinct artifacts typically appear. Low-order intermodulation distortion (IMD) and harmonic distortion (HD) occur in both sealed and leaky cabinets, sounding more subtle than the high-frequency distortion known as piano noise. Piano noise exists at frequencies higher than low-order HD and IM distortion, and studies have shown that blocking any unwanted holes or gaps can significantly improve audio quality when reproducing piano-type sounds.

[0051] It has also been found that transient multi-tone signals can highlight the piano noise problem. In some cases, increasing the number of tones excited within the high displacement bandwidth (<F0) is beneficial for enhancing distortion and improving test capabilities. Here, F0 is the resonant frequency of the driver in the enclosure. The displacement increases from DC approaching this frequency, reaching a maximum at resonance and rapidly decreasing above this frequency. The pressure in the sealed enclosure is proportional to the displacement, so the amount of the piano noise stimulus peak is proportional to the pressure and displacement.

[0052] In one embodiment, the test signal 202 can be input to the speaker as a single burst or a series of bursts. It has been found that presenting a three-tone test signal in multiple bursts (e.g., 8 bursts of three tones compared to 24 tones in a single burst) does not result in a substantial increase in amplitude, even though the density of HD and IM distortion is greater in the 24-tone signal. In one embodiment, a sparser three-tone signal that can vary over a certain excitation frequency band can be used, enabling some initial diagnostics. These initial diagnostics include using the short-time Fourier transform to identify the frequency band that excites most of the piano noise, among other diagnostics. Using the displacement for diaphragm velocity measurement as a reference curve, individual half-second bursts of three tones progressing in 1 / 3 octave steps from 200 Hz to 1 kHz are analyzed separately for the piano noise content. It can thus be confirmed that for the example speaker type, the piano noise is most prominent where the diaphragm displacement is maximum. Therefore, certain specific excitation frequency bands of an unknown speaker module can be selected for testing.

[0053] Figure 2 An example test signal 202 is shown, which has a linear decay taper within a short duration (e.g., 0.5 seconds). Regarding the signal duration, it has been found that short-duration (0.5 seconds) bursts are generally sufficient to obtain a sufficient signal-to-noise ratio to reliably detect the piano noise, so for most speakers and test conditions, increasing the cycle time is usually not necessary.

[0054] The audio signal naturally decays over time, Figure 2 An example test signal with a linear decay taper is shown. In one embodiment, different time window functions can be used to modify the test signal. For example, the test signal can have no taper to produce a rectangular window (e.g., by truncating three tones), a linear "anti-aliasing" taper, and a semi-Hann window with a 10 ms linear fade-in. The Hann window is a window function used to control spectral leakage and smooth the frequency response data. Other window functions can also be used.

[0055] Figure 7 is a graph showing the effect of windowing the stimulus amplitude with respect to time in some embodiments. Figure 7Graph 700 shows a comparison of rectangular window 702, linear and Hann windows 704, and Hann window 706 with a 10ms delay. This comparison shows that a 10ms linear fade-in (curve 706) combined with a half-Hann window produces the steepest sideband drop. Therefore, this type of window 706 is close to ideal, where the upper frequency range (above 2kHz) is empty and can be used to show even very subtle artifacts associated with piano noise.

[0056] In one embodiment, the entire upper frequency response may be left unused and dedicated to defect detection, thus allowing for easy setting of limits in quality control setups. This highlights the specific scenario of the test, where the defect is excited below 1 kHz by large diaphragm displacement / high internal enclosure pressure, and distortion can be heard as isolated high-frequency content. The test can be divided into two frequency bands, and a three-tone test signal is applied as procedural material to simulate the excitation problem and make the attenuation transients clearly audible.

[0057] Test methods

[0058] In an embodiment, Figure 2 and Figure 3 The test signal is applied to one or more loudspeakers as part of a quality control or product evaluation operation performed during production or after product distribution. The test signal may be loaded as a stimulus in the form of a WAV file (or similar file) and applied to the loudspeaker using appropriate testing equipment, where the microphone response of the loudspeaker under test can be captured to generate a response for analysis. Such a testing method can be used as part of production quality control operations or in the field by the installer or user.

[0059] The test signal provides the basis for defining normal speaker operating performance and allows users to define speaker responses that exceed normal boundaries by highlighting piano noise detected during speaker playback of the test signal. The test signal can be provided as a portable audio file that can be uploaded to a device for playback on a closed system, thus detecting manufacturing defects in a non-invasive manner.

[0060] The speaker response can be used to define a response profile or threshold that distinguishes a baseline or good speaker from a bad one. Any bad speaker can be further investigated in its frequency response within the analysis area to attempt to determine the nature and cause of the defect leading to the failure, or, as part of a pass / fail quality control check, the speaker can simply be discarded.

[0061] To test a single loudspeaker at a time, a good baseline response must be defined for comparison with the response of the loudspeaker under test. This can be obtained using historical test data derived from empirical data, or by defining a common threshold for all loudspeakers. To test a batch of loudspeakers at a time, the baseline response value can be determined by sampling all or a statistically significant number of loudspeakers and listening to their responses to derive a baseline value that is then applied to the remaining loudspeakers to be tested.

[0062] Figure 8 A method for testing a loudspeaker using a three-tone audio test signal is illustrated in some embodiments. The entire method 800 comprises three main phases: a test development phase 801, a test implementation phase 803, and a post-processing phase 805.

[0063] Test development phase 801 includes generating test signals and defining acceptable loudspeaker responses to the test signals before applying them to a batch of loudspeakers in test implementation phase 803. The first step 802 of test development phase 801 is to obtain a statistically significant number of loudspeakers to be tested. Such a number depends on the overall production batch being tested, but can be multiple loudspeakers, for example, 30+ loudspeakers. This sample is then used to obtain standard measurements, such as loudspeaker impedance, plus piano noise-specific measurements, 804.

[0064] For this process, an initial three-tone test signal is used as a burst stimulus (pseudo-piano) centered at modulus F0, 806. F0 is the resonant frequency at which piano distortion typically occurs in any loudspeaker. Previous impedance measurements from step 804 are used to pinpoint the resonant frequency. The stimulus frequency is then located there to quickly investigate the worst-case piano noise. In one embodiment, an initial batch of loudspeakers is tested to determine the limits of each of steps 804 through 808. These initial batch limits can be adjusted during initial production and then frozen for full-scale production.

[0065] In the test development phase 801, test stimuli are typically set to scan the lower frequency range of the loudspeaker to look for a series or bursts of band-increasing artifacts. Loudspeakers with high damping not only have a peak displacement at the resonant frequency but also a similar displacement at lower frequencies, making it effective to excite this low-frequency region. The system is also often tuned so that the driver sees higher voltages below resonance to boost bass, which also facilitates low-frequency stimulation and analysis.

[0066] Therefore, previous impedance measurements on the initial loudspeaker batch used to determine the maximum displacement can be used to determine the optimal test signal range and root resonant frequency. This maximizes the detection capability of the test and allows for alignment with the driver displacement.

[0067] Then, high-frequency band energy between 2kHz and 10kHz (or other analysis range) is added to generate a single dB map that distinguishes good and defective samples in terms of piano noise. Some experiments with sample modules show that bad modules typically produce artifact levels at a certain level (e.g., from 55dB to 65dB), while good modules produce lower levels of artifacts (e.g., from 40 to 43dB). This provides a relatively lenient detection margin, and the performance limit for identifying good, marginal, and bad products can be set as a basic quality control process for piano noise. The initial test signal is frequency-scalable, allowing a specific loudspeaker to be swept with the test signal at 1 / 3 octave intervals until the response contains the highest density of artifacts.

[0068] It should be noted that the dB levels mentioned here are merely illustrative examples. For embodiments that derive a single value to represent distortion, the testing process analyzes the summation of energy at discrete frequencies within a frequency band, as a reference to a specific SPL (e.g., 20 μPa) of incoherent pressure (assuming random phase). This quantifies the energy present in the analysis region, with higher values ​​indicating distortion. This method provides an effective pass / fail criterion, but it may not provide a deep frequency-discriminatory measurement, which can sometimes aid in diagnostics. However, for batch testing in production runs, pass / fail testing is efficient and cost-effective.

[0069] During the test development phase 801, technicians listen to samples and determine piano noise acceptance limits in order to classify good and bad speakers in the test batch 808. This evaluation through listening classification can then be used to set initial digital limits for speaker evaluation. Speaker responses may be subjectively categorized by listeners as good, borderline, or poor. This characterization is used to derive a single threshold that separates good (acceptable) responses from bad (unacceptable) responses. Responses representing borderline performance (where some distortion or audio artifacts are detected) can be classified as good or bad based on test tolerances, etc.

[0070] Once the test signals and appropriate good / bad profiles or thresholds are defined, they are applied to the production loudspeaker or other loudspeaker under test in the test implementation phase 803. The test implementation phase involves defining the pulse burst of the test signal (e.g., relative to the envelope, power level, decay time, etc.) Figure 2 The appropriate shape and profile of the speaker are determined, 809. Then, a properly shaped test signal is applied to each piece of software, for example, by playing a 0.5s shaped burst (e.g., fade-in + HAN) to the speaker under test using the software, 810. The resulting signal played back from the speaker is then recorded in a controlled, standardized test environment.

[0071] Post-processing stage 805 is then used to analyze the recorded results. The first post-processing step is to truncate the response into 0.5s recoded segments to eliminate noise, 811. In step 812, this process uses an FFT segment to obtain the spectrum of the noise and identify the contribution of piano distortion (HH+IMD components) to the bandwidth (above low HD). The windowed HH+IMD distortion artifacts are then summed together to obtain a single distortion SPL (sound pressure level) value, 814. This value is then compared with a defined threshold (also expressed as an SPL value) to determine whether the currently tested loudspeaker is acceptable or unacceptable. In this process, the sum of sound pressures in the analysis band is used to obtain the single distortion SPL value. A good loudspeaker will simply have a low level corresponding to the ambient noise floor, but a defective loudspeaker will have excited intermodulation tones and higher harmonic components that are fairly randomly distributed in the analysis band, such as Figure 6A The obvious difference between curves 608 (good speaker) and 610 (defective speaker) is shown.

[0072] Figure 6A An example test scenario is illustrated by applying a single 0.5s burst of test signal to a loudspeaker. In one embodiment, a series of 0.5s bursts can be applied, isolated, and then analyzed to reveal the spectral content of the loudspeaker response. This will produce a series of graphs illustrating how noticeable the increase in noise around the loudspeaker resonance is. Therefore, depending on the specific loudspeaker failure mode, the entire testing process may involve a single test signal burst or multiple bursts.

[0073] Figure 8 One embodiment is shown in which test development phase 801 uses control tests to establish frequency-related pass / fail limits that distinguish good and bad speakers. The quality control results are then used as upper limits in the analysis band. In different embodiments, pass / fail criteria can be defined by summing the energy in the analysis band and deriving a single value to signal failure. Such embodiments result in a simpler test setup but with lower resolution than the spectrum limiting methods described above.

[0074] Figure 6A Method 600 generates a single test signal for loudspeakers in samples or production batches, which may be represented as microtransducer loudspeakers with defined sizes and configurations or size / configuration ranges. Such a test signal divides the spectrum into stimulus and analysis bands to provide easy detection of distortion artifacts (identifying one as a cause and the other as a consequence), while defining a three-tone interval and level to avoid overlapping excitations and overemphasis on certain frequencies.

[0075] In one embodiment, the test signal can be scaled to suit different speaker configurations and / or test scenarios. For example, if more or less detail is needed in the data, the ratio can be halved while the resolution is doubled (1 / 6 octave), or vice versa. The start and end frequencies can also be placed within the context of a specific speaker to include its most important range. This helps control the amount of data collected regarding more or less detail. For different speaker sizes, the test scheme can simply initiate a tri-tone stimulus at a higher or lower frequency to suit the speaker under test and ensure it passes through the resonant frequency to excite as many artifacts as possible.

[0076] Therefore, this embodiment describes an audio loudspeaker testing method that uses a three-tone test signal to test piano noise distortion, where the distortion artifacts are audible and measurable. This is used to set a threshold for acceptable performance based on hearing assessments. The test signal contains stimuli that are spectrally separate from the distortion artifacts, facilitating the setting of quality control limits and data collection. The test signal is characterized by three selectable formant tones, similar to piano chords, such that frequencies and harmonics remain distributed across the frequency range. The three formant tones are selected to excite intermodulation (IMD) artifacts and also to excite harmonic distortion (HD) artifacts. The test signal is configured with a specific time window to simulate the piano time envelope and minimize spectral leakage, thereby providing sufficient measurement margin. The shaped test signal is then applied to any other loudspeaker under test to provide pass / fail analysis of these loudspeakers regarding excessive piano noise issues.

[0077] In one embodiment, the test method may be provided or used as part of the test system. Figure 9 A test system is shown that uses a three-tone test signal to implement a test method in some embodiments. For example... Figure 9As shown, system 100 includes a test signal generator 902 that generates a test signal having a piano chord-like tone with three selectable formant peaks, such that frequencies and harmonics are maintained at a frequency distribution, wherein the test signal occupies the stimulation region within the audio spectrum, and is applied to a loudspeaker. The test signal is input to one or more loudspeakers 906 under test. The test signal can be derived by first providing the test signal to a reference loudspeaker to obtain baseline measurements of standard audio response and piano noise distortion. The test signal applied to the loudspeaker 906 under test can be a pulse signal shaped with respect to power, decay time, and decay envelope shape to provide a basis for identifying the piano noise component within the frequency response of the loudspeaker 906. The output of the loudspeaker 906 is picked up by a microphone 910 and provided to an analyzer 908, which measures the microphone response of the loudspeaker to capture the frequency response and compares the measured microphone response to a baseline value derived from the reference loudspeaker 904 to determine whether the loudspeaker 906 performs well or poorly. A graphical user interface (GUI) 912 is then used to report the results to a user.

[0078] System 900 is provided for illustrative purposes only, and other components and subcomponents may be used or substituted for those shown. These components are configured to provide at least some hardware elements for execution. Figure 8 The test method 800 includes all or at least some of the processing steps.

[0079] Unless the context explicitly requires it, throughout the specification and claims, the words “comprising,” “including,” etc., should be interpreted in an inclusive sense, not an exclusive or exhaustive one. Words used in the singular or plural form also include the plural or singular, respectively. When the word “or” is used to refer to a list of two or more items, the word encompasses all of the following interpretations: any item in the list, all items in the list, and any combination of items in the list.

[0080] While one or more implementations have been described by way of example and according to specific embodiments, it should be understood that the one or more implementations are not limited thereto. In cases where more than one embodiment is described in detail, any important features common to the multiple embodiments are indeed common to those embodiments. This specification is intended to cover various modifications and similar arrangements that will be obvious to those skilled in the art. Therefore, the scope of the appended claims should be given the broadest interpretation to include all such modifications and similar arrangements.

Claims

1. A method for testing a loudspeaker to detect piano noise distortion reproduced by the loudspeaker, comprising: A transient test signal with three different formant tones is generated, the formant tones being configured to allow frequencies and harmonics to remain distributed across frequencies, wherein the formant tones are adapted to excite intermodulation distortion (IMD) artifacts and harmonic distortion (HD) artifacts from a loudspeaker, and wherein the test signal is configured with a specific time window to simulate a piano-type time envelope and minimize spectral leakage, and wherein the test signal is a pulse signal and is contained within a stimulus region of the audio spectrum; Apply the test signal to the speaker; The microphone response of the loudspeaker is measured to capture the frequency response of the audio spectrum within an analysis region separate from the stimulus region; as well as The measured microphone response is compared with the derived baseline value to determine whether the speaker's performance is acceptable or unacceptable, so that it can be reported to the user. The piano noise distortion mentioned above includes the IMD and HD components present in the analysis region of the test signal. The pulse signal has a sound pressure level that decays according to an attenuation envelope over a specified time period, used to obtain the frequency response of the captured loudspeaker to the test signal, and the method further includes shaping the pulse signal prior to the application step by defining at least one of the length of the time period, the initial power of the sound pressure level, and the shape of the attenuation envelope.

2. The method according to claim 1, wherein, The stimulation region is between 200 Hz and 2 kHz, and the analysis region is between 2 kHz and 10 kHz.

3. The method according to claim 1, further comprising: Obtain a statistically significant number of sample speakers from the multiple speakers to be tested; Obtain standard frequency response measurements and piano noise-specific response measurements for at least one of the sample loudspeakers; Test signals are generated from the obtained standard and piano noise-specific response measurements; Apply the test signal to other speakers in the sample speaker; Listen to the responses from each of the other speakers to deduce acceptable and unacceptable responses; as well as The sound pressure level threshold within the analysis region used to distinguish between acceptable and unacceptable responses was derived.

4. The method according to claim 1, wherein, The duration of the time interval is between 0.1 and 0.5 seconds, and the shape of the decay envelope is either a linearly decreasing decay or a windowed linear decay with time-based fade-in and fade-out.

5. The method of claim 1, further comprising summing the contributions of the IMD and HD components to obtain a single distortion value, and comparing the single distortion value with a threshold to determine whether the speaker is performing well or poorly.

6. The method of claim 1, wherein the test signal is applied to the loudspeaker as a periodic burst sequence, and wherein the response from each sequential burst is isolated and analyzed to allow examination of the spectral content of the entire loudspeaker response to reveal noise present near the resonant frequency of the loudspeaker.

7. The method according to any one of claims 1-6, wherein, The loudspeaker includes a miniature transducer, and the piano noise distortion is a type of distortion caused by air leakage from the rear of the miniature transducer to the air surrounding the loudspeaker.

8. A system for testing audio loudspeakers to detect piano noise distortion reproduced by the loudspeakers, comprising: A generator generates a transient test signal with three formant tones configured to allow frequencies and harmonics to remain distributed across frequencies, wherein the formant tones are configured to excite intermodulation distortion (IMD) artifacts and harmonic distortion (HD) artifacts from a loudspeaker, and wherein the test signal is configured with a specific time window to simulate a piano-type time envelope and minimize spectral leakage, and wherein the test signal is a pulse signal contained within a stimulus region of the audio spectrum and applied to a loudspeaker, the test signal being shaped to have a sound pressure level that decays according to an attenuation envelope over a specified time period to obtain the captured frequency response of the loudspeaker to the test signal, the shaping further defining at least one of the length of the time period, the initial power of the pulse signal, and the shape of the attenuation envelope; The analyzer measures the microphone response of the loudspeaker to capture the frequency response within an analysis region separated from the stimulus region of the audio spectrum, and compares the measured microphone response with a derived baseline value to determine whether the loudspeaker performs well or poorly, so as to report to the user. as well as A graphical user interface reports the speaker test results to the user, indicating whether the performance is acceptable or unacceptable. The piano noise distortion mentioned above includes the IMD and HD components present in the analysis region of the test signal.

9. The system according to claim 8, wherein, The analyzer also: Obtain a statistically significant number of sample speakers from the multiple speakers to be tested; Obtain standard frequency response measurements and piano noise-specific response measurements for at least one of the sample loudspeakers; Test signals are generated from the obtained standard and piano noise-specific response measurements; Apply the test signal to other speakers in the sample speaker; Listen to the responses from each of the other speakers to deduce acceptable and unacceptable responses; as well as The sound pressure level threshold within the analysis region used to distinguish between acceptable and unacceptable responses was derived.

10. A method for testing a loudspeaker, comprising: Obtain standard frequency response measurements of loudspeakers and piano noise measurements for a sample size within a batch to derive the threshold for piano noise distortion that distinguishes qualified loudspeakers from unqualified loudspeakers. A transient test signal with three formant tones, the three formant tones being similar to and selectable piano chords, such that frequencies and harmonics remain distributed across frequencies, wherein the formant tones are adapted to excite intermodulation distortion (IMD) artifacts and harmonic distortion (HD) artifacts from a loudspeaker, and wherein the test signal is configured with a specific time window to simulate a piano-type time envelope and minimize spectral leakage, and wherein the test signal is shaped into a pulse signal; The pulse signal was applied to each speaker and their respective responses were recorded. The piano noise distortion level of each speaker was derived from the recorded individual responses of each speaker. as well as The derived distortion level is compared with a threshold to determine whether each speaker performs well or poorly. The piano noise distortion described therein includes IMD and HD components present in the analysis region of the test signal, and the pulse signal is included in the stimulus region of the test signal, which is separated from the analysis region in the frequency domain. The pulse signal has a sound pressure level that decays according to an attenuation envelope over a specified time period, used to obtain the frequency response of the captured loudspeaker to the test signal, and the method further includes shaping the pulse signal prior to the application step by defining at least one of the length of the time period, the initial power of the sound pressure level, and the shape of the attenuation envelope.

11. The method according to claim 10, wherein, The duration of the time interval is between 0.1 and 0.5 seconds, and the shape of the decay envelope is either a linearly decreasing decay or a windowed linear decay with time-based fade-in and fade-out.

12. A system for testing loudspeakers, comprising: A generator is used to generate a transient test signal with three formant tones, the formant tones being similar to and selectable like piano chords, such that frequencies and harmonics remain distributed across the frequency range. The formant tones are adapted to excite intermodulation distortion (IMD) artifacts and harmonic distortion (HD) artifacts from a loudspeaker. The test signal is configured with a specific time window to simulate a piano-type time envelope and minimize spectral leakage. The test signal is shaped into a pulse signal having a sound pressure level that decays according to a decay envelope over a specified time period, for obtaining the frequency response of a captured loudspeaker to the test signal. The shape of the pulse signal defines at least one of the length of the time period, the initial power of the sound pressure level, and the shape of the decay envelope. An analyzer is used to measure the individual responses obtained by applying pulse signals to each loudspeaker, derive the piano noise distortion level of each loudspeaker from the obtained individual responses, and compare the derived distortion levels with a threshold to determine whether each loudspeaker performs adequately or not. The threshold for distinguishing adequate from inadequate loudspeakers regarding piano noise distortion is derived by obtaining standard frequency response measurements of a sample number of loudspeakers within a batch and piano noise measurement results. The piano noise distortion described therein includes IMD and HD components present in the analysis region of the test signal, and the pulse signal is included in the stimulation region of the test signal, which is separated from the analysis region in the frequency domain.

13. A non-transitory computer-readable medium storing instructions that, when executed by a processor, cause the processor to perform the method according to any one of claims 1-7 and 10-11.

14. An apparatus comprising: processor, and A non-transitory computer-readable medium storing instructions that, when executed by a processor, cause the processor to perform the method according to any one of claims 1-7 and 10-11.

15. An apparatus comprising components for performing the method according to any one of claims 1-7 and 10-11.

16. A computer program product comprising a program or instructions that, when executed by a processor, cause the processor to perform the method according to any one of claims 1-7 and 10-11.

Images