Page 1 of 1

Analyser Slope

Posted: March 26th, 2013, 17:47
by PHM
Currently white noise results in a flat spectrum curve (equal power per frequency band).
While listening to sound files with a full spectrum - full as perceived by the ear - and a high amplitude,
the low frequencies peak at the top of the analyser window - sometime higher than the peak meter itself.

I'd rather prefer it if something like pink noise would result in a quasi flat spectrum.

Attached a comparison of white noise and pink noise in resonics analyser and voxengo span.
All are peaking at -1.0 dBfs.

The curve of span theoretically needs a little offset upwards, but I think you can connect the dots anyway..

Re: Analyser Slope

Posted: March 26th, 2013, 19:27
by MaxLapierre
Yeah the high frequencies needs a boost.
I always use the log analyzer, it's way better than linear mode.

Tom, if you want inspiration for a nice spectrum analyzer, may i suggest you take a look at Winamp analyzer.
I know it's too small but I like the way the high frequencies are normalized (flat in relation to bass).

Thanks

Re: Analyser Slope

Posted: March 27th, 2013, 00:35
by Tom
PHM wrote:Currently white noise results in a flat spectrum curve (equal power per frequency band).
I'd rather prefer it if something like pink noise would result in a quasi flat spectrum.
Well, the former is the correct way of representing the spectrum and I'll likely be sticking to it for now as it matches my reference tests with most popular analyzers, including the one in SoundForge, WaveLab, and Waves PAZ.

How's SPAN set up?

See attached some quick comparisons, using SoundForge as reference. The vertical scale in Resonic Player is different, off course.
the low frequencies peak at the top of the analyser window - sometime higher than the peak meter itself.
That's true, and probably needs tweaking. The goal with the RTA in Resonic Player is to visualize what you're hearing in a musical way, while providing fairly accurate results. This should have a practical use while browsing through audio files. However, I'll be refining this further in future releases.

As I already mentioned Resonic Pro will have an accurate mode, with all the vertical scaling disabled and more options. We could also think about different scaling options here...
MaxLapierre wrote:Yeah the high frequencies needs a boost.
I already boost them, as part of the musical mode. But I won't exaggerate it or it might just look wrong and end up being useless for rough analysis.
If you want inspiration for a nice spectrum analyzer, may i suggest you take a look at Winamp analyzer
Oh my, you couldn't have picked a worse example ;) It's less than useless for metering.

Re: Analyser Slope

Posted: March 27th, 2013, 12:43
by PHM
Tom wrote: Well, the former is the correct way of representing the spectrum and I'll likely be sticking to it for now as it matches my reference tests with most popular analyzers, including the one in SoundForge, WaveLab, and Waves PAZ.
I didn't know that all the other analysers use a slope of 0dB/octave and never realised that my span settings don't display true peak values of single frequencies. Enlightening :)
Tom wrote:That's true, and probably needs tweaking.
Yes, the frequency curve should be offset to true values then.
It simply doesn't make sense if one frequency peaks louder than the sum itself.
Tom wrote: The goal with the RTA in Resonic Player is to visualize what you're hearing in a musical way, while providing fairly accurate results. This should have a practical use while browsing through audio files. However, I'll be refining this further in future releases.
It's funny that you mention this. IMHO a slope of 3 or 4.5dB/oct displays the data musically and the 0dB/oct analytical (=true values).
As a matter of routine I got used to the perception that to my ears white noise sounds like high frequency hiss and therefore shouldn't result in a flat spectrum.

I would like to see a third mode with 4.5dB slope then ;)
Tom wrote:How's SPAN set up?
attachment: span settings + pink noise peaking at around -12db dBfs

Re: Analyser Slope

Posted: March 28th, 2013, 00:31
by Tom
PHM wrote:It's funny that you mention this. IMHO a slope of 3 or 4.5dB/oct displays the data musically and the 0dB/oct analytical (=true values).
Not just the scale, dynamic fall-offs as well. But, yes.
As a matter of routine I got used to the perception that to my ears white noise sounds like high frequency hiss and therefore shouldn't result in a flat spectrum.
Does actually make sense.
I would like to see a third mode with 4.5dB slope then ;)
Well, there's definitely a lot of room to play with analyzer options, adjustments, and modes... so why not. I'll be playing around with this in the Pro version, but for the time being I'll have to leave it as it is as there are more pressing issues I need to focus on. Hope you understand.

Just keep your ideas and suggestions coming, especially on the analyzer topic please!
attachment: span settings + pink noise peaking at around -12db dBfs
Aight.

Re: Analyser Slope

Posted: March 29th, 2013, 20:55
by Tom

Re: Analyser Slope

Posted: March 30th, 2013, 22:04
by PHM
In the end I wanted to express something along those lines..
From a psycoacoustic point of view, the most relevant analyzer is one that shows pink noise as flat (no slope), since this relates more closely to the auditory system, which is akin to a band of one-third octave frequency.
with the following qualification: most relevant at least to me ;)
Well, there's definitely a lot of room to play with analyzer options, adjustments, and modes... so why not. I'll be playing around with this in the Pro version, but for the time being I'll have to leave it as it is as there are more pressing issues I need to focus on. Hope you understand.
Definitely understand! Just wanted to touch upon the analyser topic.

Thank's again!

Re: Analyser Slope

Posted: March 31st, 2013, 00:01
by Tom
PHM wrote:In the end I wanted to express something along those lines..
From a psycoacoustic point of view, the most relevant analyzer is one that shows pink noise as flat (no slope), since this relates more closely to the auditory system, which is akin to a band of one-third octave frequency.
with the following qualification: most relevant at least to me ;)
Yep!

Feel free to post anything that comes to mind!

After all I'm here to make Resonic better, so any input is welcome.

Re: Analyser Slope

Posted: March 31st, 2013, 16:33
by Tom
PHM wrote:While listening to sound files with a full spectrum - full as perceived by the ear - and a high amplitude,
the low frequencies peak at the top of the analyser window - sometime higher than the peak meter itself.
Fixed this in the next alpha (8xx).

Re: Analyser Slope

Posted: September 23rd, 2013, 21:32
by wshaper
Hi
Did you ever considered Comparisonicsdisplay ?
Image

Re: Analyser Slope

Posted: September 23rd, 2013, 23:06
by Tom
wshaper wrote:Hi
Did you ever considered Comparisonicsdisplay ?
Yep, actually I've been working on a proof-of-concept about a year ago, then dropped it because it'd require reworking certain parts from scratch which I wasn't ready to do at that time, concentrating on the more essential features first. Also it'd blow up the size of the waveform cache.

However, I really want to bring this back. The original motivation behind it is wanting to visualize the bass frequencies, especially in loudness war times where your average music waveform looks like a filled rectangle ;)

Re: Analyser Slope

Posted: September 23rd, 2013, 23:15
by wshaper
yipee !

Re: Analyser Slope

Posted: April 12th, 2014, 18:37
by ExtremRaym
Comparisonics on the to-do list.... AWESOME !!

(+1 for Analyser Slope improvements)

Re: Analyser Slope

Posted: April 12th, 2014, 19:28
by wshaper
Nice.

Re: Analyser Slope

Posted: April 13th, 2014, 14:23
by Tom
Well, I wouldn't be using Comparisonics though, which would be off the table anyways - licensing cost and all.

Guess I'll revisit my own code some day. See "fluid waveform" #007, #008, #012 @ http://resonic.at/labs

Re: Analyser Slope

Posted: April 13th, 2014, 14:40
by ExtremRaym
So bad for Comparisonics but I understand... and your fluid waveform" #007, #008, #012... Waoh :) GREAT !!!

I want it in my DAW now ! :D