The second sentence: "The average id dB is therefore 40 dB". You do not calculate an average of dB values like that. The average of 20dB and 60dB is 57dB.

This depends on the calculated analysis. For the spontaneous delta cursor the standard single value is always calculated in the diagram. In the case of an averaged FFT, for example, this is the quadratic sum of the levels of the included frequencies. If other single values are required, these can simply be selected in the analysis settings. The values are then calculated directly for each curve. They can be displayed for each curve in the diagram or put in a list directly (excel).

1.: Yes, the procedure is correct. If you want to know the damping for each resonance, this can be done using the averaged FFT. The overall damping can be determined from the time signal. 2.: The advantage of the hammer test lies in the ability to calculate a Frequency Response Function (FRF) using controlled excitation, where both input and output are known. A curve fitter can then process the measured FRF and accurately calculate resonance and damping for each mode.

@HEADacousticsInternational Thanks for the response, brother. 1: Okay, totally got it. 2: Without the hammer, we are getting the time vs. frequency data and then converting it to FFT in OriginLab, as there is only one output response. Now, with the hammer impact, we will have input data for the impact (force) and the output data we are already getting. How can we combine them into one FFT? Is there any method for this? Also the FRF would be only for force( input) or for both?

The FRF H(f) takes into account both input and output. There are multiple ways to calculate it. First, the FRF can be approximated by direct division of your input spectra I(f) and output spectra O(f): H(f) ≈ O(f)/I(f). This can however lead to instable results under certain conditions, e.g. when the input spectrum is close to zero for specific frequencies. The robust and accurate way is calculating the FRF based on the auto and cross spectra of input and output. The cross spectrum of input and output is: S_IO(f) = I(f) * O_conj(f), where O_conj is the complex conjugated output spectrum. The input auto spectrum is: S_II(f) = I(f) * I_conj(f), where I_conj is the complex conjugated input spectrum. Division of the two yields the FRF: H(f) = S_IO(f)/S_II(f). By the way, if you want to give this a try in ArtemiS SUITE, you might want to consider applying for our modal trial: www.head-acoustics.com/modal-analysis . All you need is your recorded time data, and we can provide the required software modules. 30 days free of charge.

@@HEADacousticsInternational How to request a 30 days free trial?

Thank you for your feedback. Feel free to check the second part and ask for a free trial license. You will be amazed.

Thanks for your tipp. That' s a good suggestion. 👍

done 😀

We don't make the videos for the likes, but to help our customers in their daily work. But yes, every👍 is motivating and even more 😀.

That is where his passion lies. Football is just about earning some extra money.

@@HEADacousticsInternational 🤣 great content in any case!

Das freut uns zu hören. Wir legen weiter nach. Am besten den Kanal abonnieren 😀

Es gibt zurzeit so viele Themen, dass wir uns zunächst auf die Vermittlung in der Weltsprache Englisch konzentrieren. Würde Ihnen zusätzlich zum Englischen auch ein deutscher Untertitel in KZbin genügen?

and that was the problem for a long time. If you wanted to work with colleagues on sound quality. an averaged sound pressure level was often the only basis for discussion. Thanks to this analysis in combination with filtered playback, you can suddenly work in a very targeted way and be understood immediately - by everyone :-)

Very insightful question! You are correct that headphone devices have an inherent advantage for this metric: There is no (or only very little) crosstalk between the channels and the entire setup is symmetrical. This does not mean, however, that all headphones achieve identical scores for this metric. For example, different frequency regions have a different impact on our spatial perception. This is taken into consideration in the binaural hearing model and will change the correlograms depending on the frequency response of the device.

thank you for your reply @@HEADacousticsInternational . Is the "ideal" spatial perception FR based on listening tests like some of your other data, or is there a particular baseline FR that you've found to objectively test better for spacial perception?

We’re getting fairly deep into the technical details here. 😊 The correlogram analysis is carried out in multiple frequency bands in parallel. In order to make the resulting information accessible for the next stages in MDAQS, we combine all of the subband correlograms into one broadband correlogram. Within this combination step, all of the subband correlograms are weighted depending on the center frequency of the respective subband. The weighting coefficients are based on earlier research results about human spatial perception. Thus, there is no explicit comparison of the frequency response of the device with an ideal spatial perception frequency response - the frequency response of the device is only relevant for the correlogram analysis because it changes the spectrum of the signal. You are already very familiar with the topics. Instead of discussing your specific technical questions publicly here, we suggest that you simply contact us by email. Then we can discuss your questions in a private conversation Regards [email protected]

seems like that 😀

The software has a modular structure and can therefore be adapted to the needs of each customer. Please send us an email to [email protected] describing your measurement and analysis tasks. We can then put together the right package for you. Greetings from HEAD acoustics

The device in this setup displays the current measured value - not an average value. The reason for the 50 dB is that ambient noise such as the ventilation system is humming quietly at a low level. People quickly get used to this and block it out in their perception. The decibel value does not do this. The dB(A) weighting is often used for such quiet background noises. This would reduce the level in the frequency range by a good 35 dB. If we set the device to dB(A), it would display 20 dB(A) as the current quiet level. Measuring 0 dB is hardly achievable in practice. We recommend using the loudness and many irritations are gone. Please have a look at the following video part 4. There these effects are explained in detail.

Yes, that's weird. That's why we did the math again with the calculator 😀. 1+1=4 is also surprising

Hello, that is correct. Service is very important to us. Every customer receives individual advice and support for their applications. Simply send an email to [email protected] with the information that you are interested in tonality analysis. Welcome on board Your HEAD acoustics team

Very insightful question! Our approach was closely related to your assumption: We indeed looked at the frequency responses of the audio systems that performed best in our listening tests. But instead of simply averaging the curves, we identified common features of these frequency responses: Which frequency regions are amplified? Which are attenuated? How smooth is the curve? Then designed a target frequency response based on these features.

Hello Okan - long time - no see 😀. When talking about „addition of dB”, we usually want to know the resulting Sound Pressure Level (SPL) in dB, when two sound sources (with individual SPL) are summed. So we are in fact interested in the SPL of the sum of the sound pressure waves. However, the sound pressure waves should not be confused with the root mean square (RMS) sound pressure, which is used for calculation of sound pressure. To avoid confusion let’s use some variables: s1(t) is the sound pressure wave of the first sound, p1 is the RMS sound pressure and L1 is the SPL in dB. s2(t) is the sound pressure wave of the second sound, p2 is the RMS sound pressure and L2 is the SPL in dB. Note that L1 is calculated from p1 as L1 = 20*log10(p1/p0) (p0 is the reference sound pressure) Lets call this the equation for SPL using sound pressure. This is mathematically equivalent to calculating L1 = 10*log10(p1^2/p0^2). Lets call this the equation for SPL using sound power. We are now interested in the SPL L3 of the signal s3(t) = s1(t)+s(2). So this is the addition of the sound pressure waves. However, it is not correct to assume that this means that the RMS sound pressure p3 can be calculated as sum of p1 and p2. Usually we assume incoherent sources. In this case we can add the sound powers which means we can calculate p3^2 = p1^2+p2^2. With this equation we can easily calculate L3 = 10*log10(p3^2/p0^2). By substituting p3^2 by p1^2+p2^2 and solving the equation for SPL using sound power for p1^2/p0^2, we get L3 = 10*log10(10^(L1/10)+ 10^(L2/10)). This is the equation which is used in the video. So the short answer to the question: In the video we are referring to the addition of two sound pressure waves, but not addition of the RMS sound pressure of the waves. Instead we are in fact using the addition of sound power. This means, that we are assuming incoherent sources, which migth should have been mentioned in the video 🤔. This assumption is mostly valid in acoustics, if the two sounds are not emitted by the same source. The second question “So consequently, the created sound pressure level from this sound source at a given location r will also be increased as 3dB once the another source is added; is that right?” can not be answered without knowledge of the spatial positions of the sound sources and the definition of the position of r. If the two sources are close to each other and we are looking at a distance r in far field (where the two sources can be interpreted as one point source), the assumption is correct.

Thank you for your feedback. There are also devices that can do both and calculate the sone and decibel values for each measurement. We show such a device, for example, from the second half of this tutorial on calculating decibel values. kzbin.info/www/bejne/an3EqY2bl81qiNE Both loud noise and low-noise microphones can be connected here to correctly evaluate every kind of noises in decibels or sone.

Then it fits. We are a German company 😀

You are welcome!

What sensors do you use?

Yes, you can. There is an extra "Edit" area in ArtemiS SUITE for this purpose. Here you can use a tool to cut out exactly one moment in time during a measurement, for example. Even better is the option of simply erasing the interfering frequencies with the mouse in the FFT vs. time analysis (as if it were an image in Photoshop). ArtemiS SUITE then generates the appropriate time-variant filters, which change the measurement in such a way that the noise is actually removed from the measurement and an FFT of the signal then looks the same (reverse engineering). The amazing thing is that there is no crackling and no drop in level. This has already saved many a measurement for our customers.

​@@HEADacousticsInternational Could you elaborate, where to find this tool and how many seconds can you cut out? I was able to create a HDF for the part before and after the noise, but the merge tool will only allow to merge files that are no longer than 2 seconds apart. I use ArtimiS Suite 7.3 and have recordings of traffic noise (drive by) 2 hours long. In these 2 hours there are parts that I would like to cut out like a plane flying by or a tractor drive by (20s- 60 s). Could you remove these parts from the HDF file and create a continuous HDF just without the tractors, noise ect ? I do not want to filter.

As an ArtemiS user, you are welcome to contact HEAD acoustics support if you have any questions regarding operation. Simply send an email to [email protected] We have already forwarded your questions to the support team. Please contact them with the reference "Cutting with AS 7" in the subject line. This way, your applications and questions will remain confidential.

Just write an email to [email protected] and we find the best option for your application. cu

Great question - and one that highlights some of the potential pitfalls of “marketing” and potentially non-standardized testing and reporting. Your products state an average noise reduction of x dB below 1kHz, but they do not state: 1) What testing methods or technologies were used? a. Did they use a state-of-the-art background noise reproduction system (compliant to ETSI TS 103 224) and a Head and Torso Simulator with ITU-T P.57 4.4 ears? b. What background noise source material was used? (Pink noise? Airplane cabin noise? Pub/cafeteria noise? Etc.) 2) How the results were derived? Are they just calculating overall levels and subtracting them? Are they subtracting spectra and then levels? 3) What is the frequency based attenuation? a. Where does peak attenuation occur? If peak attenuation for one device occurs at 50Hz, vs. say, 200Hz for another device, the same dB value for attenuation will be perceived very differently - simply because the human hearing system is not as sensitive to noise at 50Hz. b. How steady/stable is the attenuation across frequencies <1kHz? c. What is the attenuation above 1kHz!? This matters as well from a perceptual standpoint, even if it isn’t much impacted by the ACTIVE noise cancellation of the device. (>1kHz is generally dominated by the passive components) Using HEAD acoustics tools and techniques can answer all those questions and leave no doubt about the noise cancellation performance of your device(s). Hope that helps - and appreciate you watching the video! 🙏

@@HEADacousticsInternational Thank you for the answer, I really appreciate that. I also liked your collaboration with Verge regarding ANC headphones. I am especially glad that you released technical report from ANC measurements that accompanied Verge video. I hope that in the future you will be releasing more of this kind of consumer friendly reports about ANC headphones, they really make a difference.

We do not advertise headphones - We develop the proper equipment to test them 😊 Here’s a video from The Verge who have used our equipment, our facilities and our support to answer your question: kzbin.info/www/bejne/Y6uoYnapgpKLgqc

Thanks for your nice feedback. There is more about to come :-)

Hi Rebecca, here is the good news. The graphical determination of loudness over time is only asked during the study to train the understanding. In professional life, you just select the measurement and the ISO 532-1 in the ArtemiS SUITE and click "calculate" 😀

The use of the VISOR in an engine test bench is one of the most common applications. Noises of the high-pressure pump, valves, belt run-out and run-in, etc.... All sources are cleanly separated from each other and their occurrence is shown and evaluated in relation to the crank angle. If you are interested in such sample measurements, please contact us at [email protected]. Then we will be happy to show you what is possible. So far, all engine engineers have been enthusiastic about this. 🤗

@@HEADacousticsInternational I am very interested in such things but I’d need to see proof it you can actually pick up such precise sound locations.

Did you already have your live demo at your site to get the proof on your test objects? Hence we have just released a new version of the HEAD VISOR. Smaller, ligther, easier to operate and easier to afford. 😁 Just write an email to us so we can meet at your company. It is powerful and fun. cu

Hi Laxmi, thank you for your message. We will forward your greetings to Manjula. 👍

Hello, yes that is a good hint. On the measurement data offered for download here, this simple drag&drop selection of measurements works, because as explained, these are measurements taken by an SQuadriga - a frontend from HEAD acoustics. These frontends write natively into the hdf-format directly . In case you want to evaluate data from third-party devices in ArtemiS, we provide extensive import options. These are actually at your fingertips with a right mouse click. 😀 Thanks for your hint.

We like to read that. You are welcome.

Our software requires either transfer functions for the individual surface points to determine the modal parameters. Or you use the time signals from the occurring motions to calculate an operational deflection shape (ODS). We would like to talk about this in more detail. To get in personal touch, please mail to [email protected]

Hello M. Nebra, we run our tests in an anonymized and confidential testing environment and unfortunately cannot disclose any details about the specific models being tested. This is also necessary to ensure fair competition. We hope you understand.

@@HEADacousticsInternational sure, it's understandable, of course. It's simply curiosity. That would be a very useful information as a consumer, but I can see the potential usability of this test by the audio gear companies. This test is revolutionary. Thank you for your answer.

Hello disienna, we run our tests in an anonymized and confidential testing environment and unfortunately cannot disclose any details about the specific models being tested. This is also necessary to ensure fair competition. We hope you understand.

@@HEADacousticsInternational I was curious because I found a similar curve to sound ideal about a decade ago that I formulated with homemade rigs (my head and a homemade ear canal) and my home theater that I calibrated using known small room psychoacoustics. It’s sort of cool that this experiment found an exceptionally similar curve. I’m also glad that this supports some other suspicions I’ve had regarding audio quality relating to headphones. Anyway, I’d love to own this headphone just to hear it, but I guess that is impossible. I’ve searched long and hard.

Thanks for your detailed feedback. Feel invited to watch the other three videos in this series as well. Then you will have a good overview of the handling of acoustic measurement and analysis methods. You can find the links in the description box (More...)

We're glad we could help you get started. We have many more videos here that will help you overcome the typical difficulties in understanding modal analysis. Enjoy!