Wishlist (from current version 3.9.8.1)

[RJA4000]

Hello

I'm using MI intensively right now, and therefore, here I am again with a Wishlist.
You've done such a great job in the past to improve the tool based on our feedback...
Thanks again for that.

So here is my new list.

1. Analyze from file
As listed in another post, don't reset anything on analysis from WAV file except Oscilloscope/probe

2. Multitone Frequency accuracy
Currently, it's not possible to enter frequency with fraction.
And even if one is entered in the .TCF file, it will be dropped on reading.
This is very annoying, since it doesn't allow to avoid spectral leakage with Multitone.

3. Multitone Spectral leakage
On top of the above, an option to avoid spectral leakage for each periodical Tone in the multitone would be very handy.
And if the flags are saved in the file, this could then even be applied at playback only:
Then one single Multitone definition could be used with varied Sampling Frequency/FFT size and always work.
Say you save the file with a Sine at 997Hz with the "Avoid spectral leakage" option set, and when you reload the file, the frequency is adapted according to the current SF and FFT Size.
If that's going too far, a single checkbox in the Generator and apply the option for all perdiodic tones when playing the tones would still work.

5. Dithering
Proper dithering is critical for audio measurements.
As you explained, the MI generator doesn't dither signals.
You suggest to use Multitone, but currently, given point above, it's unpractical.
Anyway, it would be much beneficial to have a direct easy way to add dithering in a controlled way in the generator:
And to allow to select bit depth, function, strength,... even, if possible (although much less cirtical) separately for each channel

Also, or maybe alternatively, a way to simply add dither (with the above-described level of details) in the Multitone would be very welcome.
(If the point 2 above is solved, of course)
But it won't solve all cases (read point 11 below)

6. Phase for each channel in generator
The generator allows setting phase difference between the 2 channels
But there is no way to directly set the overall phase for both.
(Again, if yous except the use of the Multitone tool)

7. Spectrum analysis: Energy in defined Bands
This is a very handy feature, and I use it a lot.
Unfortunately, there is no way to enter the list of the frequency bands quickly:
One has to enter each of them separately, which may be very long.
It would be VERY helpful to be able to load a list from a CSV file (with enough accuracy to allow selecting one single bin) or to paste it from Clipboard.
Or both.
Then I could use Excel to prepare the list and copy/paste it easily.
(By the way: Why not to allow entering just a start frequency and a number of FFT bins ? OK, that one is very optional)
(Oh, and I just figured out that I need to specify frequencies [bin #]-0.5 to [bin #]+0.5 to get all of one bin's energy.
Would be great to add that to the manual, IMO)

8. Spectrum analysis: Peaks detection
Mutlitone (32 tones) analysis is very helpful for Audio measurements.
I use AudioPrecision generated 32 tone WAV files.
But it always misses the 25Hz tone, for some reason, even in Rectangle FFT window with perfect sync.
I wonder if there would be a way to give more control on the selected frequencies.
OK, if the above point 7 is sorted, I could also do all that manually.
Would have to re-compute TD+N myself though.

9. International support - Decimal separator
That would be very handy to have a way to change some localisation settings.
Specifically the decimal separator (we use "," here)
By reading windows settings or just allowing a choice in an options menu, like "Display" Menu.
The number of times I screwed by typing a wrong Output Level amplitude...
Where it's important: In generator. Also in Spectrum analyzer User Defined Frequency Bands.
Also in export CSV format (Where a Tab separator option could also come handy)

10. Generator Output amplitude in dBFS
I asked that in the past, didn't I ?

11. Generator WAV file replay is not bit perfect
I don't know why, but when playing back RME's bit test files, it always fails.
Only it gives 24 bits bit perfect message if I play the 24 bits test file... in 32 bits mode in MI.
Do you add dithering ? Or change the level ?
I don't know, but that's strange.
Also, see above comment about dithering from the generator: It would be helpful to control addition of dithering when a WAV file is played at a reduced volume.
One more vote for more direct dithering control in the generator.

12. Allow (even) more DDPs
I know you've jumped from 16 to 24 DDPs. Thanks 1000 times for that.
But what about 32 ?

13. Spectrum Analyiser - More direct control on vertical and horizontal scroll bars position
It would be handy to have a way to enter directly a min value, as an example, rather than having to play with the mouse to position the scale where we want to see it.

14. Spectrum analyzer - Please add 180dB range to Spectrum analyzer
(Granted, it's not the most urgent.)

15. Spectrum Analysis - Energy in user defined frequency bands - Frequency accuracy
It seems that the accuracy of the frequency is decreasing with the frequency:
For a 48kHz sampling frequency with 64k FFT Size,
From 10Hz to <100Hz, it uses 4 digits (0.0001) accuracy, leading to an average error of 0.0%
From 100Hz to <1000Hz, it uses 3 digits (0.001) accuracy, leading to an average error of 0.1%
From 1000Hz to <10000Hz, it uses 2 digits (0.01) accuracy, leading to an average error of 0.7%
From 10000Hz to <100000Hz, it uses 1 digits (0.1) accuracy, leading to an average error of 6.8%
I understand it keeps the same number of significant digits (6).
But the FFT bin width doesn't change with frequency, and therefore, the last 1 (or 2) ranges become(s) realy too much to get accurate results in some cases.
Would that be possible to always store frequencies with 0.001 (max) resolution (ie to keep 8 significants digits instead of 6) ?

16. Hide channel B from spectrum and Oscilloscope
I use Channel B to record additional values (typically: RMS Volt level) to display on screen.
But if channel A+B are selected, both channels are displayed in Spectrum anaylisi and Oscilloscope views.
Of course, for Spectrum, that's easy: I may just change the vertical scale and send the B plot to the sky (or hell).
But if I want to do the same for oscilloscope, then I'll have to switch off "Auto scaling" for the ADC... which is very unfortunate.
Would it be possible to have a "Hide B" option for Spectrum and Oscilloscope ?
OK, that one is "nice to have". But nice, it would be.


I've colored in red the most important topics for me.

[VirtinsTech]

Thank you for all the suggestions.

1. This function will be provided in the coming new version MI V3.9.9.0. See viewtopic.php?f=8&t=3614 for detail.

2, 3, 5 see viewtopic.php?f=8&t=3613&p=5019#p5019

6. yes, Multitone should be used if the phase needs to be specified.

7. If you repeatedly use the same setting, you can save it as a Panel Setting File. If you want to read data from individual FFT Bins, you can probably use Spectrum Analyzer Export. We might add the suggested CSV file loading/saving in the future.

8. Increasing Record Length and FFT size will increase the peak detection resolution, which is determined by FFT resolution and the width of main lobe of the window function used.

9. Probably no. "," can cause confusion in a CSV file. Also, "Copy As Text" provides Tab separated format.

10. Already provided. See Section 4.3.3 of the software manual.

11. see the link in the above 2. To avoid resampling and rescaling the DAC sampling rate, bit depth, amplitude of the signal on the signal generator must be the same as the WAV file in order to try to keep it bit perfect.

12. Maybe in the future.

13. Probably no.

14. Probably no.

15. Finest resolution 0.00001Hz will be allowed for any frequency band definition in the coming new version MI 3.9.9.0.

16. To hide Ch.B display, set "B Range" at the bottom of the screen to "OFF", or simply right click a graph window and select "Display A ony"

[RJA4000]

Thanks for taking the time to answer !

1. This function will be provided in the coming new version MI V3.9.9.0. See viewtopic.php?f=8&t=3614 for detail.

GREAT ! Thanks

2, 3, 5 see viewtopic.php?f=8&t=3613&p=5019#p5019

The fact Multitone doesn't allow fractional frequencies (and phases) is a MAJOR drawback.
I answered in the other topic for more details.

6. yes, Multitone should be used if the phase needs to be specified.

BUT... I can't create a signal without spectral leakage then... (See the point before.)

7. If you repeatedly use the same setting, you can save it as a Panel Setting File. If you want to read data from individual FFT Bins, you can probably use Spectrum Analyzer Export. We might add the suggested CSV file loading/saving in the future.

Yes, of course. That's what I currently do.
But then I'll end up with a huge amount of PSF files.
I can't just figure up upfront what exact settings I'll need for all cases.
And creating the list currently takes me dozens of minutes for each PSF.
So the addition of the loading of CSV file, at least, would be a BIG added value, indeed.

8. Increasing Record Length and FFT size will increase the peak detection resolution, which is determined by FFT resolution and the width of main lobe of the window function used.

Well, sure.
But still, I don't get hwo this works.
Here is an example:

MT32 - Peak detection - FFT Crop.png (34.2 KiB) Viewed 18162 times

MT32 - Peak detection - Frequencies.png (28.91 KiB) Viewed 18162 times

MT32 - Peak detection - Settings.png (27.55 KiB) Viewed 18162 times

(You may find the multi tone definition in the topic you linked to 2,3,5)

9. Probably no. "," can cause confusion in a CSV file. Also, "Copy As Text" provides Tab separated format.

In our countries, separator is ";" instead of ",".
But OK, don't worry about this one.

10. Already provided. See Section 4.3.3 of the software manual.

I know I can see it in dBFS and I know I could use the verticall slider to set it when dBFS is selected.
But that doesn't (completely) solve my issue: Most of the time (always), I want to set it to an exact value.
Granted, the slider is quite accurate for values close to 0dBFS.
OK. It's for sure not a priority.

11. see the link in the above 2. To avoid resampling and rescaling the DAC sampling rate, bit depth, amplitude of the signal on the signal generator must be the same as the WAV file in order to try to keep it bit perfect.

Which is the case in my example.
I have a workaround, so it's not very important.

12. Maybe in the future.

13. Probably no.
14. Probably no.

One may always ask...

15. Finest resolution 0.00001Hz will be allowed for any frequency band definition in the coming new version MI 3.9.9.0.

That will be a great improvement.
Thanks

16. To hide Ch.B display, set "B Range" at the bottom of the screen to "OFF", or simply right click a graph window and select "Display A ony"

This is the moment I feel dump...
So you're even better than I thought.

[RJA4000]

Just to highlight what I'm trying to do with Multi Instrument:
https://www.audiosciencereview.com/forum/index.php?threads/review-and-measurements-of-topping-e30-ii-lite-dac.47518/

[VirtinsTech]

Thanks for your posting on AudioScienceReview.

Well, sure.
But still, I don't get how this works.
Here is an example:

The resolution of the peak detection function in MI is defined by the Dead Band and the width of the main lobe of the window function used, whichever is wider. The role of the main lobe is not mentioned in the current software manual. It will be updated.

The main lobe width used in MI is usually equal to or greater than 15 x [frequency resolution] depending on the Window function used. If rectangle window is used here, then the finest resolution of the peak detection would be 15 x [frequency resolution] = 15 x 0.732422Hz = 11 Hz. This explains why 25Hz is missing between 20.508Hz and 32.227Hz. Increasing Record Length and FFT Size should solve the problem.

[RJA4000]

Thanks for the explanation.

Indeed, it makes sense to add it to the manual. That will be more obvious to everyone.

[RJA4000]

17. DDP for Sampling rate, FFT size, Number of points, Number of channels

18. For UDDP functions, add ROUND(x,y) and INT(x)

[VirtinsTech]

7.

Yes, of course. That's what I currently do.
But then I'll end up with a huge amount of PSF files.
I can't just figure up upfront what exact settings I'll need for all cases.
And creating the list currently takes me dozens of minutes for each PSF.
So the addition of the loading of CSV file, at least, would be a BIG added value, indeed.

Save and Load as a CSV file will be provided in the coming V 3.9.9.0.

12. Maximum number of DDP viewers will be increased from 24 to 32 in the coming V 3.9.9.0.

Any good reasons for items 17 & 18?

[RJA4000]

Thanks !

About 17 and 18, I think you know the answer.
To compute bin start and end, we need Round function, Sampling frequency and FFT size.