Sound pressure / loudness of impulse noise

[run...@gmail.com]

First of all, thank you four your useful software! I would like to be able to measure impulse noise regarding its loudness. And while I know that impulse noise can be too loud for correct measurement I would feel the following features useful. Thank you very much.

Alternative Impulse-time weighting

(like fast and slow) - standard is 25 ms I guess. Variable value would be even nicer. This wouldn't conceal the true loudness of an impulse as much as 125 ms, or am I thinking wrong. There is still one problem. As fta can't be updated that often it would be available as a single value only?

LC(peak)

peak sound pressure level, from Wikipedia: Using either 'C' or 'Z' frequency weighting. 'Peak sound pressure level' should not be confused with 'MAX sound pressure level'. 'Max sound pressure level' is the highest RMS reading a conventional sound level meter gives over a stated period for a given time-weighting (S, F, or I) and can be many decibel less than the peak value

[Julian Bunn]

Hi,

Thanks for these interesting suggestions. I think the Impulse response timing is now realistic to implement (I've held off on it since the outset), since the frame time (i.e. speed at which each set of samples is processed) on most modern Android devices is 10ms or less, which makes a decay calculation of duration 25ms sensible. So, I will add that to my feature list to implement.

I'm not clear why LC(peak) is needed given that AudioTool measures LEQ for which there are Fast/Medium/Slow filters that can be set?

Julian

--
You received this message because you are subscribed to the Google Groups "AudioTool Discussion Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email to audiotool-discussio...@googlegroups.com.
To post to this group, send email to audiotool-dis...@googlegroups.com.
Visit this group at https://groups.google.com/group/audiotool-discussion-group.
For more options, visit https://groups.google.com/d/optout.

[run...@gmail.com]

Glad to hear that you liked my suggestions.

Just to make sure I understand this correctly:

Samples are taken very often every second (e.g. 44 kHz). After a certain amount of samples fft is used to calculate the spectrum. Many different features use this spectrum (octave-analysis, peak, full res.). Single values could be calculated more often (like SPL), but are bound to the fft at the moment. As fft took very long in the past only 125 ms were a realistic decay time. No value in the software is calculated more often than every 125 ms. There is a realtime sound graph, but SPL is only calculated every 125 ms.

As it is now possible to fun fft in 10 ms (I can confirm this from my device), It would be possible now to lower decay time to 25 ms. 

Now to your question:

As you sure know: 'Peak sound pressure level' should not be confused with 'MAX sound pressure level'. LEQ is often used as an indicator for inner ear biological stress. My interest regarding impulse noise is also in this direction. Everyday noise consists of impulses too (banging car doors etc.) If there are really sharp impulses like clapping I guess than the calculations (even fast) wouldn't take that into account. So global impulse time weighing would improve LEQ too.
Impulse noise has a different effect on the ear than normal noise as the protective mechanisms of the ear are worse for very short noise (like a gun shoot). A 1.5 ms noise with 160 dB at the ear can damage it permanent to some degree even thought LEQ wouldn't be influenced by it very much, when I understand it right.
LC(peak) for that reason is necessary to measure for workers protection which is a personal interest of me. It records the loudest point in the sound wave. I guess it needs real time analysis of the sound wave like in that oscilloscope view in the app. Am I right that until now it already refreshes at 10 ms?

Thank you very much

[Julian Bunn]

I'm currently out of town and need more time to reply ... I'll try to do so in the next couple of weeks.

[Julian Bunn]

Sorry for delay in replying ... some answers/comments below:

On Thu, Jun 30, 2016 at 1:55 PM, <run...@gmail.com> wrote:

Glad to hear that you liked my suggestions.

Just to make sure I understand this correctly:

Samples are taken very often every second (e.g. 44 kHz). After a certain amount of samples fft is used to calculate the spectrum. Many different features use this spectrum (octave-analysis, peak, full res.). Single values could be calculated more often (like SPL), but are bound to the fft at the moment. As fft took very long in the past only 125 ms were a realistic decay time. No value in the software is calculated more often than every 125 ms. There is a realtime sound graph, but SPL is only calculated every 125 ms.

The displayed SPL is exponentially adjusted with a time constant to fit the user selected speed. In the "Fast" setting this is 0.125 seconds, in "Medium" 0.5 seconds, and "Slow" 1.0 seconds. Since many phones can accumulate sample buffers and FFT them at intervals of 0.01 seconds or less, then an "Impulse" speed could be useful, which would be a 0.025 second​ decay time, as you suggest below.

The interval at which the phone can retrieve data is determined by the sample buffer size being used and the sample rate. 

 

As it is now possible to fun fft in 10 ms (I can confirm this from my device), It would be possible now to lower decay time to 25 ms. 

Now to your question:

As you sure know: 'Peak sound pressure level' should not be confused with 'MAX sound pressure level'. LEQ is often used as an indicator for inner ear biological stress. My interest regarding impulse noise is also in this direction. Everyday noise consists of impulses too (banging car doors etc.) If there are really sharp impulses like clapping I guess than the calculations (even fast) wouldn't take that into account. So global impulse time weighing would improve LEQ too.
Impulse noise has a different effect on the ear than normal noise as the protective mechanisms of the ear are worse for very short noise (like a gun shoot). A 1.5 ms noise with 160 dB at the ear can damage it permanent to some degree even thought LEQ wouldn't be influenced by it very much, when I understand it right.
LC(peak) for that reason is necessary to measure for workers protection which is a personal interest of me. It records the loudest point in the sound wave. I guess it needs real time analysis of the sound wave like in that oscilloscope view in the app. Am I right that until now it already refreshes at 10 ms?

​Yes, that display is refreshed for every new buffer received from the microphone. Typically, for a 4096 length buffer at 44.1kHz rate this is ​every 0.1 seconds (4096/44100).

It would be possible to add a peak ADC level display to the Mic (Oscilloscope) view, if that would be useful?

Julian

 

Thank you very much

--
You received this message because you are subscribed to the Google Groups "AudioTool Discussion Group" group.

To unsubscribe from this group and stop receiving emails from it, send an email to audiotool-discussion-group+unsub...@googlegroups.com.
To post to this group, send email to audiotool-discussion-group@googlegroups.com.

[run...@gmail.com]

I think the peak value would be helpful. That's the max. Amplitude right?

Will think about it further an can hopefully edit my answer.

[run...@gmail.com]

I have already found your update of 25 ms peak decay time very helpful. It really uses more calculation power but on my Samsung s4 mini it is still usable. Will you also implement the oscilloscope peak feature you suggested? What determines the buffer size? Why is such a buffer used?

Thanks a lot!

I have rated your app 5 stars.

[Julian Bunn]

Thanks for the rating.

The buffer is used for the FFT calculations. The longer the buffer, the more precise the frequency resolution, but the longer it takes to collect the samples, and the longer the FFT takes to compute. At 44,100 sample/sec data rate, a 44,100 length buffer would only update the display every second, which would feel painfully slow, but would have sub Hz resolution in frequency.

So, the length of the buffer is a compromise, really. 

There is an option in the Menu to increase the FFT size, called "Fine Resolution".

The peak feature on the Mic display is on my to-do list :-)

Julian

[Non Timebo Mala]

Your work is very valuable and much appreciated.

I am interested in pursuing this line of thinking a bit further.

I am working specifically with airgun moderators.  I've recorded many,  many shots with another tool (the Parrot app) and pulled them into Audacity for analysis.  The impulse which I am looking at (trying to isolate) has a rise time on the order of 0.1 ms and a duration of 1 cycle with a period of about 2 ms (so between 4 kHz and 6 kHz).  I don't care what the frequency actually is, I just want to get an SPL on that "pop".   It takes really high dollar equipment to collect something of that duration principally because such hardware does a whole lot of other things that I don't need or want.  Do you think a buffer size of 256 or 512 is even possible?  That would give you a ~6 ms window and would aid people working niche problems like this greatly?

Now I realize the implementation of that might be more trouble than it is worth or even unworkable so I have a second question, please.  I you were trying to measure the SPL of that signal with this software what settings would you optimize?

Thank you for your time sir.

Five stars on you.

On Monday, October 3, 2016 at 4:27:28 PM UTC-4 Julian Bunn wrote:

Thanks for the rating.

The buffer is used for the FFT calculations. The longer the buffer, the more precise the frequency resolution, but the longer it takes to collect the samples, and the longer the FFT takes to compute. At 44,100 sample/sec data rate, a 44,100 length buffer would only update the display every second, which would feel painfully slow, but would have sub Hz resolution in frequency.

So, the length of the buffer is a compromise, really. 

There is an option in the Menu to increase the FFT size, called "Fine Resolution".

The peak feature on the Mic display is on my to-do list :-)

Julian

On Mon, Oct 3, 2016 at 1:14 PM, <run...@gmail.com> wrote:

I have already found your update of 25 ms peak decay time very helpful. It really uses more calculation power but on my Samsung s4 mini it is still usable. Will you also implement the oscilloscope peak feature you suggested? What determines the buffer size? Why is such a buffer used?

Thanks a lot!

I have rated your app 5 stars.

--
You received this message because you are subscribed to the Google Groups "AudioTool Discussion Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email to audiotool-discussion-group+unsub...@googlegroups.com.

To post to this group, send email to audiotool-dis...@googlegroups.com.

[Julian Bunn]

Lower sample sizes would be possible, but they are of limited use to most people, I think.

For measuring the SPL of this signal, I would set AudioTool to the Impulse (fastest) metering, and then set the Peak Hold on. Then, after the signal, you could use the cursors to examine the SPL of the peak observed in the spectrum (and its frequency). I'm not sure how effective this would be?

Julian

To unsubscribe from this group and stop receiving emails from it, send an email to audiotool-discussio...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/audiotool-discussion-group/b1cdae9e-3a54-47f3-9e39-c198a99fbb36n%40googlegroups.com.