multitone spectrum proposal

[solderdude]

First you need to describe the effect 'grungy' as it is a subjective term that can mean just about anything to anyone.
Grungy might possibly related to limiters and dynamic compression for loudness effects in which case it is unrecoverable.
Maybe grungy is similar to 'coarseness' in sound ?
Grunge is a music genre so perhaps 'grungy' would be a better term than 'grunge' to describe a specific sound effect ?
Second can you post a short clip of a recording with the 'grungy sound' and one that doesn't have this ?
The thing is that mid frequencies are about the easiest to record and reproduce well.

 

[KSTR]

As music is not about average levels but more about impulses and peak levels (which multitone looks like as a signal) I found the 'average level' research to be close to the green line I drew in below.
But in the research for peak levels (which is what music is more about) I found a similar slope but about 20dB higher in level and a decade higher in frequency.

Well, a typical multitone like AP's looks like what it is, a sum of constant level sines with "random" start phases, phases choosen to achieve low crest factor.
It differs from music that it does not contain "blips" (high energy narrow-band short signal bursts -- think of shaped envelope sine burst of a few cycles).
A way to come close to this is to use beating, each frequency of the multitone replaced by a doublet 1 Hz or so apart. This generates an amplitude modulation at 2Hz rate without polluting the spectrum.

Generally, I don't like dense multitones with 32 equally log-spaced frequencies as they completely pollute the noise floor with all the IMD products. I strongly prefer sparser multitone with frequencies carefully selected so that there is enough room on the bottom to show the noise floor, with the IMD components clearly sticking out.

As for peak levels vs. frequency, this of course depends on that you use as music signal. With modern rock/pop you will find that it contains peak levels in all frequency groups at almost the same level, only sloping down at the extreme frequency ends. This can be shown by using 1/3rd octave filters (linear phase, preferably) feeding peak hold meters. The RMS averages vs frequency of course still follow pink'ish slopes of 10dB/dec.

 

[LesOrr]

First you need to describe the effect 'grungy' as it is a subjective term that can mean just about anything to anyone.
Grungy might possibly related to limiters and dynamic compression for loudness effects in which case it is unrecoverable.
Maybe grungy is similar to 'coarseness' in sound ?
Grunge is a music genre so perhaps 'grungy' would be a better term than 'grunge' to describe a specific sound effect ?
Second can you post a short clip of a recording with the 'grungy sound' and one that doesn't have this ?
The thing is that mid frequencies are about the easiest to record and reproduce well.

I agree with all of your points. I especially acknowkedge that 'effects' employed by the recording engineer might make the recovery of a reasonable version of the original 'live' sound impossible.

Speech intelligibility is my 'pet peave' more so than music reproduction. That said, a very good example of a muddy grungy almost uninteligible recording is the sound track from the recent movie "Elvis", especially the opening scenes.

BTW, "Grunge" is Jensen's word - I prefer "Mud." His presentation is quite detailed, including the test equipment, the math, spectral analysis and conclusions. It is Copyrighted and available from the AES site for a small fee. He used several test signals >= 10 kHz to cause IMD in the DUT and measured the resulting distortion products mainly below that.

 

[LesOrr]

Well, a typical multitone like AP's looks like what it is, a sum of constant level sines with "random" start phases ...

KSTR: I use an AP SYS 1 including a burst card, a Tektronix SG5010 + AA5001 (with homemede Python code) and a Crown IMD. I have not figured out how get them to perform the multitone tests that you referred to. Is there a thread on that or a related subject somewhere? It would be helpful not to need to design and construct a test fixture to perform these tests.

 

[LesOrr]

... A way to come close to this is to use beating, each frequency of the multitone replaced by a doublet 1 Hz or so apart. This generates an amplitude modulation at 2Hz rate without polluting the spectrum.

Generally, I don't like dense multitones with 32 equally log-spaced frequencies as they completely pollute the noise floor with all the IMD products. I strongly prefer sparser multitone with frequencies carefully selected so that there is enough room on the bottom to show the noise floor, with the IMD components clearly sticking out. ...

KSTR: An afterthought - I also have + use 2 Tektronix SG505s. I was thinking of replicating the sine wave generator circuit in those several times and mixing/summing the results to generate the multitone test signal. That still leaves me with the problem of notching out the fundametals after the DUT to separate out the resulting distortion residual.

 

[solderdude]

You can generate multitones (and lots more) with the (free) DeltaWave software from @pkane.

 

[LesOrr]

WOW ! Thank you again. You opened up an entire group of info I couldn't find by myself. I really appreciate it.

 

[pkane]

You can generate multitones (and lots more) with the (free) DeltaWave software from @pkane.

Using Multitone Analyzer pink spectrum generator:

 

[LesOrr]

Pkane: Thank you. I have not pursued digital measurements because I do not have an instrumentation quality ADC or DAC. Is there one that you would recommend? Your results are a lot more useful than any I have acheived digitally.

 

[Multicore]

I know my proposal is not a standard and maybe even nonsensical but here it goes.

Given how often I have read Amir's review comments to his own 20-tone graphs remarking on its unrealistic energy distribution relative to that of music, this seems interesting.

 

[solderdude]

Pkane: Thank you. I have not pursued digital measurements because I do not have an instrumentation quality ADC or DAC. Is there one that you would recommend? Your results are a lot more useful than any I have acheived digitally.

If you want affordable and measurement quality I can recommend this:

E1DA Cosmos ADCiso Review

This is a review and detailed measurements of the E1DAC Cosmos ADCiso specialized Analog to Digital Converter (ADC). The grade "B" was kindly purchased by a member and costs US $179. As you see, you have dual balanced inputs. Back side has two USB-C inputs: one for power and one for computer...

www.audiosciencereview.com

You can make it more versatile using the scaler:

Cosmos Scaler | index

e1dashz.wixsite.com

 

[markstein]

Oh, and that is also already being done - even though much too rarely.

Attached is at the left a multitone measurement of a QSC CP8 speaker by the German magazine Production Partner (embedded for research purposes; fair use clause applies), taken by Prof. Dr. Anselm Goertz with his Monkey Forest software. To the right is my measurement of the same speaker under the same circumstances with ARTA. The frequency shaping applied was EIA-426B (green curve); the resulting frequency response in red (Production Partner) and purple (mine); and the distortion products below that.

View attachment 435723 View attachment 435724

I really would like too see this multitone test done in each speaker review.

Speaker tests by Anselm Goertz in Production Partner or Sound & Recording are truly exceptional.

 

[Sokel]

Pkane: Thank you. I have not pursued digital measurements because I do not have an instrumentation quality ADC or DAC. Is there one that you would recommend? Your results are a lot more useful than any I have acheived digitally.

The above measurement suite is free, its author its @pkane (who posted the test) and it has his own thread here:

Beta Test: Multitone Loopback Analyzer software

This is a software beta-test for IMD and THD aficionados with an intense desire for multi-tone audio testing :) The start of the documentation and the download website is here: https://distortaudio.org/multitone.html (for 64-bit Windows) Multi-tone Analyzer (MA) was more of an experiment than...

www.audiosciencereview.com

along with @solderdude 's suggestion you can achieve really good results.

 

[KSTR]

KSTR: An afterthought - I also have + use 2 Tektronix SG505s. I was thinking of replicating the sine wave generator circuit in those several times and mixing/summing the results to generate the multitone test signal. That still leaves me with the problem of notching out the fundametals after the DUT to separate out the resulting distortion residual.

I see, you like good stuff ;-)
My SG505 has the 60Hz IMD option that I've used occasionally... Not a very pure sine, though, restricting its usefulness. And the main Oscillator is plagued by mains residuals induced from the local transformer.

These days, a good soundcard and REW or Multitone give better results overall and are much more versatile most of the time.. I also have written a few quick&dirty C proggies that generate various signals.

 

[Hayabusa]

That was my point..

Testing equipment at the peak and to well known standards is a good thing.

About that 'peak':
A 32 multi tone has a peak to average ratio of around 13dB.
This means these peaks at 0 dB are not really 0 dB.
So then my next question:
Why are multi tone plots not shifted to the right level?
The noise floor is now also displayed at the wrong level?
Maybe @amirm can answer this?

 

[KSTR]

About that 'peak':
A 32 multi tone has a peak to average ratio of around 13dB.
This means these peaks at 0 dB are not really 0 dB.
So then my next question:
Why are multi tone plots not shifted to the right revel?
The noise floor is now also displayed at the wrong level?
Maybe @amirm can answer this?

I agree this is an issue with regard to noise floor reading. OTOH, the noise floor is already skewed by "FFT gain".

I think Multitones should always be fully specified, notably the resulting crest factor or even better yet, the frequencies (or FFT bin numbers) and start phases of the sine or cosine oscillators and if they are periodic then the block size.

I can create one and the same dense multitone with regard to frequencies and magnitude spectrum with varying crest factors from close to infinity (basically almost a Dirac pulse) to almost unity (endless sine sweep, overlapping the range ends). Of course they give totally different result even when normalized to the same RMS level.

 

[amirm]

Maybe @amirm can answer this?

I always shift the graph to be at zero dB up top. Is this what you are asking?

 

[Hayabusa]

I always shift the graph to be at zero dB up top. Is this what you are asking?

Yes, the peaks are at a different level, may be better to leave them at (I assume) -13dB or explain with what level they are shifted?

 

[LesOrr]

What I am taking away is that most of the preceding responses equate digital testing with analog testing. If there is a similar thread from an only analog testing perspective, I would really appreciate a referral to that url.

 

[Lambda]

I always shift the graph to be at zero dB up top. Is this what you are asking?

Why?
To me this sounds misleading.

As chaining the scale also changes the "noise floor"