Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: VST which promises to fight the loudness war (Read 37878 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

VST which promises to fight the loudness war

Reply #25 –
Can I ask a dumb question?

Why did anyone think this made an audible difference?!

People are reporting huge audible differences, but once you level match (which I'll admit is tricky), there's... nothing.


I suppose it could have a real effect on certain down-stream processing.

Cheers,
David.
 

VST which promises to fight the loudness war

Reply #26 –
Let me show you one cool trick with ReLife. Run some audio file through ReLife, and then reverse the result in time and ReLife again. You'll get your original audio file (minus some fixed gain).

I've just figured out that ReLife is an all-pass (i.e. phase-shifing) filter, it does not do any change to the frequency response. It shifts phase at low frequencies.

VST which promises to fight the loudness war

Reply #27 –
Quote from: 2Bdecided on
Can I ask a dumb question?

Why did anyone think this made an audible difference?!

People are reporting huge audible differences, but once you level match (which I'll admit is tricky), there's... nothing.
Actually... I made a DBT which resulted in hearing differences. I used your replaygain for it to bring both files to the same level. Was that the wrong way? If so, I could have been fooled by different levels. Or was I responsing to phase variances? I was under the believe that they can´t be heard.
marlene-d.blogspot.com

VST which promises to fight the loudness war

Reply #28 –
Quote from: Cavaille on
I used your replaygain for it to bring both files to the same level. Was that the wrong way?
It's probably one "right" way, but it's not perfect. It would need any level differences due to ReplayGain's inaccuracy to be un-ABXable, and I'd never claim that!

If something introduces a fixed amplitude change, a phase change, and no frequency response change, then it's best to measure the level change with a sine wave. If something introduces a signal-dependent amplitude change, together with a phase change, then it's really difficult to calculate the level change with real music.

I'll tell you how I did it once: I got the levels as close as I could, then changed the level of one signal in 0.5dB steps, then 0.1dB steps, going backwards and forwards (like tuning an analogue radio) to find the spot where the levels sounded closest. I used ABX at that level, and 0.1dB either side of it, and then 0.2dB either side of it.

If there's just a level difference, then (assuming you got close to start with!) one of those levels should be un-ABXable. If there's more than a level difference, you'll be able to ABX it at all levels.

Hope this makes sense. It doesn't sound very scientific, but at the end of the day, human ears are best at judging how loud things sound to human ears - with changes like we're dealing with here, I don't think automatic measurements can be trusted (yet). Or if there's some that can, it's because they've been proven to match human ears!

Cheers,
David.

VST which promises to fight the loudness war

Reply #29 –
Quote from: Alexey Lukin on
Let me show you one cool trick with ReLife. Run some audio file through ReLife, and then reverse the result in time and ReLife again. You'll get your original audio file (minus some fixed gain).


I did it with peak algo 3, and that's the frequency response of result:


VST which promises to fight the loudness war

Reply #30 –
Isn't that to be expected? When you compress frequencies, the levels are lower... or what?
Can't wait for a HD-AAC encoder :P

VST which promises to fight the loudness war

Reply #31 –
This frequency response deviation is only 0.1 dB and its effect is insignificant.
Most of the visible effect is produced by phase response variation.

VST which promises to fight the loudness war

Reply #32 –
Quote from: Alexey Lukin on
Let me show you one cool trick with ReLife. Run some audio file through ReLife, and then reverse the result in time and ReLife again. You'll get your original audio file (minus some fixed gain).

I've just figured out that ReLife is an all-pass (i.e. phase-shifing) filter, it does not do any change to the frequency response. It shifts phase at low frequencies.

Thanks for that discovery, Alexey! Took me a bit to understand how an all-pass can "de-clip" a waveform, but now I got it. As said, the plug-in doesn't clean up the spectrum of the clipped waveform (i.e. remove distortions), it just hides the clipping by breaking the phase coherence of clipped sections. That's probably fine for people who believe an MP3 sounds bad because "the spectrum above 16 kHz is missing" ("it looks better, so i know it will sound better"). I favor the iZotope approach towards signal processing

Chris
If I don't reply to your reply, it means I agree with you.

VST which promises to fight the loudness war

Reply #33 –
While the positive 'psycho-visual' effect of this kind of phase shifting has shown to be out of question, can the possibility of a positive psycho-acoustic effect on compressed material really already be ruled out?

VST which promises to fight the loudness war

Reply #34 –
Quote from: 2Bdecided on
It's probably one "right" way, but it's not perfect. It would need any level differences due to ReplayGain's inaccuracy to be un-ABXable, and I'd never claim that!
(...)
Hope this makes sense. It doesn't sound very scientific, but at the end of the day, human ears are best at judging how loud things sound to human ears - with changes like we're dealing with here, I don't think automatic measurements can be trusted (yet). Or if there's some that can, it's because they've been proven to match human ears!
Man, that sounds time consuming! But it makes sense. I´ll maybe try it that way. And David, thank you very much for the detailed reply.


Quote from: C.R.Helmrich on
Thanks for that discovery, Alexey! Took me a bit to understand how an all-pass can "de-clip" a waveform, but now I got it. As said, the plug-in doesn't clean up the spectrum of the clipped waveform (i.e. remove distortions), it just hides the clipping by breaking the phase coherence of clipped sections. That's probably fine for people who believe an MP3 sounds bad because "the spectrum above 16 kHz is missing" ("it looks better, so i know it will sound better"). I favor the iZotope approach towards signal processing
I favor the iZotope approach also - but one can combine those two which works quite nicely considering the fact that one re-synthesize and the other one "lies". I did it on some files and the results were... stunning.

BTW, even iZotope can be "fooled": the histogram inside the DeClipper shows a signal with a perfectly natural dynamic response with a "ReLived" file. Before it showed only the imperfect response with the usual "bumps".


Quote from: googlebot on
While the positive 'psycho-visual' effect of this kind of phase shifting has shown to be out of question, can the possibility of a positive psycho-acoustic effect on compressed material really already be ruled out?
Yes, and that´s IMO the most important question. Can we rule this possibility out? As written above it is fairly time consuming to really figure out if this thing actually creates a sonic improvement for some people. However, with the methods given to me right now (Replaygain, DBT etc.) I concluded for myself that ReLife creates a sonical more pleasing audio signal when compared to the original, compressed signal. And since we don´t usually have the uncompressed signal... for some people this may simply be too ridiculous to consider - I mean, who wants to introduce phase distortions into an audio signal on purpose? Well, I always read about people preferring some old EQ or some old plug-in because it "sounds" nice while in reality it just adds another error. And of course the people who are not interested in technical explanations... like the usual snake-oil customers.

I decided for myself that this tool isn´t that bad... I mean, the signal on which it´s used is already destroyed and far from being "pure" so does it really hurt to "impure" it even more? But this is another discussion because I still don´t know why phase variations / distortions are bad. I´ve googled it but only found technical descriptions on solutions to prevent them. I still don´t know why they need to be prevented. Are they hurting the loudspeakers? Are they introducing some nonlinear distortions? If I reverse the phase will that mean that the loudspeaker driver will suck instead of push (well that makes sense - but what if it only reverses the phase slightly? Is that possible?).

Since I don´t know it - yet - I´ll continue to use it. I don´t mind a little bit of psycho-acoustic cosmetics... for me, the end justifies the means.

P.S.: this is me talking... Until now I avoided any analogue modelled thing in my DAWs - and now this! 
marlene-d.blogspot.com

VST which promises to fight the loudness war

Reply #35 –
Quote from: Cavaille on
Since I don´t know it - yet - I´ll continue to use it. I don´t mind a little bit of psycho-acoustic cosmetics... for me, the end justifies the means.


Well, according to Mr.Lukin's finding, ReLife doesn't really do anything to the source material other than skewing the phase response. (I seriously doubt it'd be audible) If you combine ReLife with iZotope, you only get iZotope with ~4dB cut & phase shift. I would rather combine iZotope with an upward expander or a ratio adjustable dynamic compressor. I find the result from iZotope alone is still too timid, and the upward expander boosts everything that hits the threshold to the same level; since peaks are all normalzed on compressed music, upward expander does not work great alone either. So if you only expand, or midly 'exaggerate' what iZotope has synthesized, you might be able to get an ideal result.

BTW I figured Flux Bittersweet II works quite well too (and it's FREE!) It's too bad the plugin lacks an adjustable threshold control though.. :$