REAPER Contest - June 2017 (RESULTS)

[karbomusic]

Originally Posted by REAmix

Can't argue with that. But if people aren't meeting the requirements, then their tracks should be normalized, just like real life when you send your file to Spotify or YouTube, and it might not come out how you'd hoped.

If they don't follow the conditions, and their tracks aren't changed, and if that gives them an edge and gets them votes, it would be the opposite of teaching them to follow the criteria, and would penalize the people that did.

I have no problem with that or rejecting submissions until they get it right - just like in the real world. Like I said, I didn't read all the replies leading up to mine so I could have missed something context wise.

[DaveKeehl]

Congratulations to Reno.thestraws for winning the Reaper Contest June edition

Sir, please choose a song for the next month from here and let me know with a PM

Thanks everybody for sticking by even during summer time! You're the best!

PS: Sorry Fergler, it looks you're too good even with the bass muted! ahahahahaha

Download all the submissions

[Tod]

Humm, I just checked out the SWS Loundness tool and it's pretty cool. However, the peak values it shows are very strange from what I can tell.

I tried it with 6 tracks I had rendered to put on CD, and they all had peak levels at -0.6dB except on one song. Actually I rendered them at
-0.5dB but Reaper alway shows everything 0.1dB less, at least that's been my experience.

It showed all the tracks as being above 0.0dB starting at 0.1dBTP going up from there. What's really strange, is that the only track that was actually above -0.6dB was the one that showed 0.1dBTP which is the lowest in the readings.

I'm much more concerned about the peak values then I am the LUFS, I just want them to be fairly even on the CD.

I think I trust the Orban analyzer more, and when I have some time I'll need to run them all through that to compare the LUFS reading in SWS.

[REAmix]

could it be interpeak samples somehow giving you the anomolous readings? I'm not sure I understood properly what you explained, but I would imagine that Orban, and hopefully the Normalizer worked with Trupeaks. But maybe the Normalizer doesn't?

[karbomusic]

It appears dbTP is a calculated best guess and may/will display differently in different tools I've found. That being the case, I'd probably just pick a tool and stick with it for consistency. I use Orban because it was easy to use, does a queue of files offline and can be pointed to any channels on the sound card etc. with a couple handy options and I got used to that convenience. I like measuring the files offline since that's literally the final result so to speak.

[Tod]

Originally Posted by REAmix

could it be interpeak samples somehow giving you the anomolous readings. I'm not sure I understood properly what you explained, but I would imagine that Orban, and hopefully the Normalizer worked with Trupeaks. But maybe the Normalizer doesn't?

Hi REAmix, no I don't think so, I think interpeak samples create much smaller differences and are related more to peaks that are near 0.0dB where they become most important, at least that's my knowledge of them.

In all honesty, I'm not sure what true peaks really is. I can conjure up things in my mind but the readings I get just don't add up. I'm probably missing something here too.

At any rate, I would be careful about using the SWS Loundness tool to make adjustments to the audio files.

I'm anxious to check the LUFS out against Orban, if SWS is good there then it's still a useful tool. But I still wouldn't use it for adjusting the audio files unless there's a better understanding of the peak levels.

[karbomusic]

Originally Posted by Tod

In all honesty, I'm not sure what true peaks really is. I can conjure up things in my mind but the readings I get just don't add up. I'm probably missing something here too.

Try this...



Think of the black dots as samples and with those we can reconstruct that sine wave there. But samples are just data values and if you normalize (for example) to zero dBFS that's where the dots are going to be placed; not the top of the sine wave because it doesn't actually exist, it's just numbers to be used in order to recreate it.

Now when that normalized set of samples gets 'reconstructed' into the original sine wave, the result now extends above zero (the dotted portion of the sine wave), but zero is as far as we can go. Those are intersample peaks which is what dbTP is trying to reconstruct and measure because they are the true peaks not the sample value peaks.

So when we try to normalize to say -0.5 dBFS, we are just trying to leave enough room to account for those, or go by the dbTP number which tries to help out in figuring how much wiggle room is required.

[martinmadero]

Originally Posted by DaveKeehl

Congratulations to Reno.thestraws for winning the Reaper Contest June edition

Sir, please choose a song for the next month from here and let me know with a PM

Thanks everybody for sticking by even during summer time! You're the best!

PS: Sorry Fergler, it looks you're too good even with the bass muted! ahahahahaha

Download all the submissions

Congrats to the winner!!! Reno.thestraws
and all who participate in this month!!!

[MacFizz]

Well, well
I thought my mix sounded pretty good...
At least I have identified a problem with my monitors. Too much treebles in there...
See ya for next month contest ?!

[DaveKeehl]

Originally Posted by MacFizz

Well, well
I thought my mix sounded pretty good...
At least I have identified a problem with my monitors. Too much treebles in there...
See ya for next month contest ?!

Sorry MacFizz I forgot to put your name in the rankings below! You could try to use reference mixes to check your work if you have monitoring issues Hope I see next month!

[msundh]

This is what I miss. Feedback.
I thought my mix was good last month, but got no votes. Was it really bad or was I everybody's number 4?
What happened to the discussion we had earlier about changing the voting?

[Lunganar]

Originally Posted by msundh

This is what I miss. Feedback.
I thought my mix was good last month, but got no votes. Was it really bad or was I everybody's number 4?
What happened to the discussion we had earlier about changing the voting?

I know the feeling, maybe we should rank all submissions instead of only the 3 best. The winner has to choose a new track, and the lowest ranked 'wins' some mix advice by the rest of the forum so that he actually learns something for the next mix.
Only the best and the worst projects are shared. One to learn from, and one to advice on.

[karbomusic]

Originally Posted by msundh

This is what I miss. Feedback.

I used to encourage everyone to list each submission by ID and give their thoughts about them, which I always did myself. Usually once someone takes that first step, others follow.

[martinmadero]

here´s some feedback.
I think this month, especially and without exception, no mix sounded decent
I think there were many votes for discarding comparing which was the least bad and not highlighting the qualities of each particular mix
I personally listen to the mixes with the monitors I use to mix and with the headphones and none convinced me
The votes in any case already commented that they are a separate topic and may or may not reflect technical criteria or simply of taste
For me it is essential that a good mix is ​​balanced, that no instrument stands out from the other and is heard as a whole, something that I could not hear here, personally in the balance and relationship between drums / bass
Personally among my votes was the winner, chose the mix for equalization and the balance of vocals and guitar solo.
The mix of fergler was among my favorites, but I still can not convince myself. Seeing your project I found too much compression.
My mix I did very fast and I escaped details, I left the guitar solo under which it was already late to change it. I focus too much on the drums, and forget many things.
The intention to have chosen this track was to sound well something that was originally badly recorded. Is very easy and comfortable when the tracks and production are perfect since every detail we add makes it sound better, but in the case of these live recordings, it was a real challenge.

[Tod]

Originally Posted by karbomusic

Try this...



Think of the black dots as samples and with those we can reconstruct that sine wave there. But samples are just data values and if you normalize (for example) to zero dBFS that's where the dots are going to be placed; not the top of the sine wave because it doesn't actually exist, it's just numbers to be used in order to recreate it.

Now when that normalized set of samples gets 'reconstructed' into the original sine wave, the result now extends above zero (the dotted portion of the sine wave), but zero is as far as we can go. Those are intersample peaks which is what dbTP is trying to reconstruct and measure because they are the true peaks not the sample value peaks.

So when we try to normalize to say -0.5 dBFS, we are just trying to leave enough room to account for those, or go by the dbTP number which tries to help out in figuring how much wiggle room is required.

Yeah, that's a very good example Karbo, and that's what I considered as interpeak samples.

That's why I can't understand what the "True Peak" readings in the "SWS Loudness" are supposed signify, I think they're way beyond what interpeak samples might do. I might add, 1 out of the 6 tracks was over -0.6dB and it shows the least of all the True Peaks in the SWS action.

I tried to post this picture but I haven't been able to post pictures or links for the last 3 days. It shows the peak readings for 6 tracks along with the SWS Loudnes tool.



I'm just posting this to make everybody aware that the "SWS Loudness Tool" might not be totally accurate.

If I get some time today, I'm going to check the LUFS in Orban, to see how they correlate with SWS Loudness.

[karbomusic]

Originally Posted by Tod

[img]/31030/SWS%20Loudness%20Tool%201.png[/img]

The image link is missing the prefix and domain. Should be... http://somedomain.suffix/31030/SWS%2...20Tool%201.png or https://somedomain.suffix/31030/SWS%...20Tool%201.png Where somedomain.suffix is just what I made up to show the pattern.

[Tod]

Originally Posted by karbomusic

The image link is missing the prefix and domain. Should be... http://somedomain.suffix/31030/SWS%2...20Tool%201.png or https://somedomain.suffix/31030/SWS%...20Tool%201.png Where somedomain.suffix is just what I made up to show the pattern.

Oh yeah, thanks Karbo, I should have noticed that. I just tried to add the "http:/" then "https:/", but it still didn't take, there must be something else too. So it's the "stash" that's screwing up.

At any rate, I just tested the 6 songs in Orban and the LUFS is pretty close, one was exactly the same and the rest were .1 LUFS off. For example what SWS showed as -12.5 LUFS is -12.4 LUFS in Orban.

Incidentally, has anyone else tried to us "stash" in the last 3 days?

[bezusheist]

here...



bbq code help page...https://forum.cockos.com/misc.php?do=bbcode#imgcode

[Tod]

Originally Posted by bezusheist

here...



bbq code help page...https://forum.cockos.com/misc.php?do=bbcode#imgcode

Aah, thanks bezusheist, you've rescued me twice in the last 2 days.

I wonder if someone should mention to Justin whats happening in stash, or maybe I'm the only one.

[karbomusic]

Originally Posted by Tod

I wonder if someone should mention to Justin whats happening in stash, or maybe I'm the only one.

Not sure if it was a typo but you referenced 1 slash / not two // earlier. You need both just after "http:". Below is what the entire proper image link looks like, minus the [IMG] tags...

https://stash.reaper.fm/31030/SWS%20Loudness%20Tool%201.png

Btw, I don't know what method you guys use to get the image url, I've always done it manually so excuse my not knowing the more automatic way if that's where the problem happens to be.

[Tod]

Originally Posted by karbomusic

Not sure if it was a typo but you referenced 1 slash / not two // earlier. You need both just after "http:". Below is what the entire proper image link looks like, minus the [IMG] tags...

https://stash.reaper.fm/31030/SWS%20Loudness%20Tool%201.png

Btw, I don't know what method you guys use to get the image url, I've always done it manually so excuse my not knowing the more automatic way if that's where the problem happens to be.

Yeah thanks Karbo, that's basically what bezusheist pointed out.

I've always just copied the links provided in stash, but lately they are totally incomplete.

Do you use stash at all Karbo, it's really a pretty good resource and I use it quite a bit, although I do try to be conservative with it and not take it for granted?

[SmajjL]

Ohh, thank you peoples! and bezusheist, after all this time, there was a link on how to do the youtube one even..
If that have been there for yeeeeaaars then..
10 PRINT "LoL!"
20 GOTO 10
I think

Ps, SmajjL failed today also if it helps, with stash.

[karbomusic]

Originally Posted by Tod

Do you use stash at all Karbo)

I don't actually, I use imgur.com for images, my own server (or youtube) for audio.

[Reno.thestraws]

Thanks! Just saw that "I Won"

It was a "one hour live mixing" just correcting the very bad tracking and playing

Thank you

[REAmix]

Originally Posted by karbomusic

Try this...



Think of the black dots as samples and with those we can reconstruct that sine wave there. But samples are just data values and if you normalize (for example) to zero dBFS that's where the dots are going to be placed; not the top of the sine wave because it doesn't actually exist, it's just numbers to be used in order to recreate it.

Now when that normalized set of samples gets 'reconstructed' into the original sine wave, the result now extends above zero (the dotted portion of the sine wave), but zero is as far as we can go. Those are intersample peaks which is what dbTP is trying to reconstruct and measure because they are the true peaks not the sample value peaks.

So when we try to normalize to say -0.5 dBFS, we are just trying to leave enough room to account for those, or go by the dbTP number which tries to help out in figuring how much wiggle room is required.

iZotope Ozone 7 maximizer, actually has true peak limiting on it. Ozone 7 is one hell of a suite.

[REAmix]

Originally Posted by Tod

Yeah, that's a very good example Karbo, and that's what I considered as interpeak samples.

That's why I can't understand what the "True Peak" readings in the "SWS Loudness" are supposed signify, I think they're way beyond what interpeak samples might do. I might add, 1 out of the 6 tracks was over -0.6dB and it shows the least of all the True Peaks in the SWS action.

I tried to post this picture but I haven't been able to post pictures or links for the last 3 days. It shows the peak readings for 6 tracks along with the SWS Loudnes tool.



I'm just posting this to make everybody aware that the "SWS Loudness Tool" might not be totally accurate.

If I get some time today, I'm going to check the LUFS in Orban, to see how they correlate with SWS Loudness.

The true peaks should be where the actual peaks are, once they've been reconstructed. 2.5 db certainly seems higher than one would expect, but it's not out of the norm. I've definitely gotten 2.5 db of true peaks even though sample level peaks were in the negatives.

Easy thing to do would be run the same file through Orban and see if it says the same.

I would imagine that if there is a True Peak reading there though, that it is accurate and taken into account in normalizing to a given LUFS value.

[DaveKeehl]

Originally Posted by REAmix

iZotope Ozone 7 maximizer, actually has true peak limiting on it. Ozone 7 is one hell of a suite.

yep, I use the vintage limiter all the time

[DaveKeehl]

So, to get back to the Loudness/True Peaks discussion, I would say these are the requirements from now on:

-14 LUFS, -1 TP. All the entries will be normalized to -14 LUFS to prevent big volume changes.

[Tod]

Originally Posted by REAmix

The true peaks should be where the actual peaks are, once they've been reconstructed. 2.5 db certainly seems higher than one would expect, but it's not out of the norm. I've definitely gotten 2.5 db of true peaks even though sample level peaks were in the negatives.

Easy thing to do would be run the same file through Orban and see if it says the same.

I would imagine that if there is a True Peak reading there though, that it is accurate and taken into account in normalizing to a given LUFS value.

Hi REAmix, I hadn't really paid attention, but Orban's are definitely close enough, although they call it "Highest Recontructed Peak Level", and it's either right on or has a 0.1dB difference.

Until now, or until Karbo turned me on to Orban, I had never really heard about "True Peak" before, at least not in this sense. Heh heh, I've had many hundreds of stuff go out my studio, and there's never been a problem with any of it that I know of.

So I guess I'm wondering what the significance of it is, and I'm very curious how this should influence a mix?

[karbomusic]

So I guess I'm wondering what the significance of it is, and I'm very curious how this should influence a mix?

Typically and real-world, little because a random sample here and there that makes its way through the ceiling is going to generally be inaudible and chances are in the past you were always leaving just enough room anyway so that they never even occurred. This is really a non-issue until we start trying to make all our mixes as close as possible to the highest possible value without going over.

It's also a side-effect/problem of loudness wars because when we start compressing 'that' much, we end up with far more peaks close to the ceiling that we have to deal with vs more reasonable DR with a lot less peaks that close comparatively. I suppose for all our interest here is how that occurs...

These are the same mix, the top one has the DR I normally like and use which is somewhere between -14 and -16 LUFs with peaks near the ceiling only occurring in spread out intervals. The bottom one has been limited to a more modern hyper-compressed value. There are an awful lot more peaks close to the ceiling due to compression causing more ISP calculations to deal with.



I'm sure everyone understands this, just nice to see it in a pic.

[bezusheist]

There is some good information on all of this "loudness" and "true peak" stuff at the T.C. Electronic website...
Some of the material may be a little outdated (tech is so fast these days) but most is still relevant. (Stop counting samples is a classic)
You can find all of the papers here > http://www.tcelectronic.com/loudness...ture-glossary/

[REAmix]

Originally Posted by Tod

Hi REAmix, I hadn't really paid attention, but Orban's are definitely close enough, although they call it "Highest Recontructed Peak Level", and it's either right on or has a 0.1dB difference.

Until now, or until Karbo turned me on to Orban, I had never really heard about "True Peak" before, at least not in this sense. Heh heh, I've had many hundreds of stuff go out my studio, and there's never been a problem with any of it that I know of.

So I guess I'm wondering what the significance of it is, and I'm very curious how this should influence a mix?

That's odd. If they don't readout the same, then something is wrong somewhere.

I think a small intersample peak of 1db, is relatively difficult to hear. I would have trouble hearing it on my system for sure. For a long time, I just limited to sample peaks, and I never really noticed clipping distortion.

But some other people on better systems or just with better ears, probably could.

[karbomusic]

Originally Posted by REAmix

That's odd. If they don't readout the same, then something is wrong somewhere.

With various tools, they may/won't read the same because dBTP is a bit of a calculated guess. Went through this in another thread, just load up multiple VSTs that do dBTP and watch many of them vary in their resulting dBTP number.

That makes me go back to my previous stance, if this is giving someone that much grief, they are proverbially riding too close/compressed to the ceiling to begin with. -14 shouldn't be that though.

[bezusheist]

Originally Posted by REAmix

But some other people on better systems or just with better ears, probably could.

i think it would be more of a problem for worse systems...
these "inter-sample" peaks happen in the analog domain.
only a "poorly" designed DAC would not allow for any "headroom" in the analog domain.

(edit: i know that contradicts some info i just linked to (but like i said) some is outdated...prosumer DAC technology advanced a lot in 10+ years...)

[Tod]

Originally Posted by karbomusic

Typically and real-world, little because a random sample here and there that makes its way through the ceiling is going to generally be inaudible and chances are in the past you were always leaving just enough room anyway so that they never even occurred. This is really a non-issue until we start trying to make all our mixes as close as possible to the highest possible value without going over.

It's also a side-effect/problem of loudness wars because when we start compressing 'that' much, we end up with far more peaks close to the ceiling that we have to deal with vs more reasonable DR with a lot less peaks that close comparatively. I suppose for all our interest here is how that occurs...

Thanks a lot Karbo, you basically confirmed my own thoughts, heh heh, I needed that.

Originally Posted by bezusheist

There is some good information on all of this "loudness" and "true peak" stuff at the T.C. Electronic website...
Some of the material may be a little outdated (tech is so fast these days) but most is still relevant. (Stop counting samples is a classic)
You can find all of the papers here > http://www.tcelectronic.com/loudness...ture-glossary/

That looks like a great read bezusheist, lots of AES stuf there. I bookmarked it and maybe some day when I've got more time, I can go through some of it.

Originally Posted by REAmix

That's odd. If they don't readout the same, then something is wrong somewhere.

I think a small intersample peak of 1db, is relatively difficult to hear. I would have trouble hearing it on my system for sure. For a long time, I just limited to sample peaks, and I never really noticed clipping distortion.

But some other people on better systems or just with better ears, probably could.

Actually what I was trying to convey REAmix, is that the small differences were not very significant to me. When they are that close, they're close enough to call it even.

And no, I don't think anyone can hear the peaks that interpeak sample creates. I would think the higher frequencies might be the worst culprit, and the lower freqs the least problem.

[Tod]

Something the still confuses me and don't seem logical, if you look at my picture you can see that a track (#5) that is -13.4LUFS and has a True Peak of 2.5dBTP, while a track with -ll.4LUFS has a True Peak value of 0.8dBTP? Logically to me it would be the other way around?

Then there's the track that acutally reads -0.1dB on Reapers meter, and it has the smallest True Peak value, .01dBTP?



Hummmmm???

[bezusheist]

True Peak has no correlation to LUFS, RMS, Peak, or any other scale...
It is correlated to the amount of "high frequencies", though.
IMO it's silly to use it in this contest because as we've seen, all meters give a different TP value.
So the rules here say "-1 dBTP"...but who's TP meter/reading is that referencing ?

[Tod]

Originally Posted by bezusheist

True Peak has no correlation to LUFS, RMS, Peak, or any other scale...
It is correlated to the amount of "high frequencies", though.
IMO it's silly to use it in this contest because as we've seen, all meters give a different TP value.
So the rules here say "-1 dBTP"...but who's TP meter/reading is that referencing ?

Yeah, I think we've established that True Peak is not relevant to what we're doing.

[DaveKeehl]

ok then -1 dB but not true peak

[DaveKeehl]

sorry guys if I'm not very active recently but I am so busy lately. I would love to find the time to give feedback to each user but it's impossible right now