Is the master spectrum analyzer db axis correct?

here is an image:

it shows 5 analyzers
renoise internal,
voxengo span in renoise master channel,
signalanalyzer in renoise master channel,
ableton live internal
adobe audition internal
track headroom in renoise is at 0db
the signal is the same wav in every case, a steady saw wave at 444hz
all slopes are set to 0/none
all analyzers have a block size of 16384
all analyzers display log

i dont think this is a testing method error

why is every analyzer peaking at about -6db but renoise internal at -20db ?

I tested this issue using a sine wave and compared renoise’ internal analyzer with Melda’s MAnalyzer… Renoise shows the peak at about -26dB while MAnalyzer shows it at about -20dB. One thing that you should be careful about is the slope, see if all analyzer’s slope factors are set to the same amount, preferably 0.

Try playing a sine wave above 10kHz and see how badly that gets reported!

Compare the differences between Logarithmic and Linear frequency modes. No reason these should make any difference to reported dB!

Drawing/curve types also seem to make some difference. 1/3 octave seem to be the closest to what you would expect to see…

Also just testing this and I’ve found changing Track Headroom to anything less than 6dB makes no difference! I will open a separate report for that though.

the signal is the same wav in every case, a steady saw wave at 444hz
all slopes are set to 0/none
all analyzers have a block size of 16384
all analyzers display log

i dont think this is a testing method error

Seems the higher the blocksize the less accurate the dB reported is. Especially at higher frequencies! Even with a sine wave, which should have all its energy within a single band.

actually it depends on the analyzer.with most i have used blocksize doesnt affect peak value

the other analyzers agree more or less within ±1db
only renoise internal is inaccurate.

anyway, are you suggesting that renoise master spectrum view works as it is supposed to and reports accurate db values?

No, I’m saying that Renoise’s reported levels change by an extortionate depending on the blocksize set, especially at higher frequencies. It appears to remain a lot flatter at a low blocksize but then your frequency reporting is a lot less accurate. At a high blocksize you get peaks and dips before a massive rolloff at the upper end. Doesn’t seem right at all!

i cannot reproduce what you are saying. or maybe you are just looking at log. for inspecting high frequencies, log is not ideal. try set it at linear and test again

but, this is not the point of my post/question.
i am curious about this ~15db shift in renoise analyzer

As you can see I’ve already compared Log to Lin and mentioned (some of) the differences. In fact Log and Lin seem to report different peaks and dips of level for sine wave at different frequencies.

Load the Chip-Sine instrument packages with Renoise.
Play a note at C-4 to give a nice steady tone.
Set Transpose for the instrument to 67.
In Log mode you have a measurement about the same as for the rest of the range. In Lin mode it is reported about 60dB lower.
Now try a Transpose of 66.
Now Log reports it as ridiculously low. Lin mode is also still far lower than it should be but now higher than that reported at 67 Transpose. (Finetune of -21 gives you almost the reverse display of using Transpose 67. The reading for Log falls below the bottom of the measured scale between about -5 and -20 finetune though.)

This is a good question, I’ve also noticed the levels don’t seem to be accurate when compared with some spectrum plugins I’ve used, but just assumed it was something I was doing wrong.

yes you are right. this seems to be another inaccuracy of the spectrum analyzer of renoise

Hi… I just mastered this track tonight… I think the voxengo has a little more, “hi definition,” but the metering looks the same to me…

I think the Renoise metering is accurate, but I also think that these analyzers have all kinds of settings, and readings, and graphs, and I am pretty sure that the Renoise master spectrum is looking at RMS and some of your other analyzers are looking at peak, and if you tweak there settings, you might be able to match the appearance of the one in Renoise, and therefore you can confirm for yourself and all…

Edit again = here is another image of the same track… I think it is possible that the Renoise one might even be more accurate, but if you notice with your eyes, the difference is either rs-met is not accurate, or latency is screwing it up… and the difference between the two of the is minimal imho

Below, the same track in my old copy of Live 7… That is the Spectrum Plugin from Live…

same track… same thing

hi 2 daze j, bit off topic but what theme are you using? it’s pretty pretty.

-edit-

ps, do not master at night :o

Hi phuture, its Fog by beatslaughter… Its about the 9th one down, in your themes folder. Its great, and yeah, I try not to master at night… lol

:walkman:

2 daze j, the metering in your first example looks completely different to me. it cant be right. it also peaks at completely different levels.

thats why i chose to use a constant wave so as the test would not be so much time/lag sensitive.

in your second example, renoise peaks at about -30 and signalanalyzer peaks at -18, live also peaks at -18. also these three analyzers dont have special settings about peak/rms.

Here are three more examples. The only difference in these volume numbers is: “Me being a human being, and not having a perfect eye hand coordination.”

** I am using the, “thumbnail,” setting on imgur. I did not want to add too many full size pictures, to what is becoming a busy thread. Please note the pink circles I made.

Hope this helps,

Cheers

Edit = apparently you can not click on these photos for full size counterparts. I will return with direct links… ( not with img tags )

Edit = In no particular order. Includes full sized links to the pictures in my post above, which I resized for, “forum sanity.”