|Author||Topic: Metering tools (some in-depth tests and discussion)|
Posted: 3rd May 2004 12:27
Hello dear musician colluegues.
I'm looking for some metering tools for VST or DX. Perhaps you can help me out with links and opinions. I'm expecially looking for a PEAK/RMS/VU meter (PMM isn't so important) that has not only a low latency but also shows me the overall and maximum levels and not just a short peak on the meter. It should also be very precise/correct (or as good as possible) so that I can work with the K-System (K-20, K-14, K-12). Best if it would be free or at least up to 50Euro max. The usual issue of "students don't have money".
DestroyFX RMS Buddy
This lil' devil is fast if we talk about latency and metering reaction. My negative points on it are that it has no meters and I kinda don't get the idea of averange and continual RMS indicators. The continual RMS often goes over the averange one. But I kinda need the maximum RMS peak and the actual RMS peak. Elseways it's a great plugin.
Elemental Audio INSPECTOR
...is a very great plugin. It lacks only two things: a slider for controlling the latency (actually it runs with a latency of about 300ms) and a maximum RMS indicator. I wrote Elemental Audio more than 4 times giving them new ideas for implementation and always the question when a next update will be out. This was early January and still no update. A pity for such a great plugin.
This is what I'm looking for. But I don't like the policy of purchasing a hardware key first (that costs around 60Euro itself) and then the plugin (115Euro).
Your opinions would be greatly apreaciated. Thanks in advance.
Posted: 3rd May 2004 12:34
Well, Inspector and RMS buddy are what I use, and I also refer to the RMS meter in Voxengo's Elephant mastering limiter. All three are slightly different, but careful eyeballing during the loudest sections of a given track will get you where you want to be. Once I'm fairly certain I've got a track where I want it level-wise, I export the file and then use Cool Edit to analyze the average RMS for the loudest sections of the track(s) for a fourth opinion. Usually, it's right where I want it to be (within a single dbFS or so).
Posted: 3rd May 2004 12:50
Why don't you try a real good one meter, try PSP Vintage Meter, I realtime audio metear, and more important for free.
It's at www.pspaudioware.com
Posted: 3rd May 2004 12:50
Well... if the "negative sides" on those plugins would be negotiated, then my search would be over. But still - using more than one plugin in a chain? Well, that's why I was asking. I'm kinda looking for an all on one solution.
Posted: 3rd May 2004 12:57
Honestly, if I had to rely on one plugin for RMS level checking, it'd be Elephant, since it's pretty much always at the end of my master bus anyway, and it seems as reliable as the other two. However, I don't know if you want to buy a mastering limiter just to get RMS metering (even though Elephant is, in my opinion, an essential plugin purchase). Also, the way it does metering is very similar to RMS Buddy, which you don't like, so... sorry I couldn't be of more help!
Posted: 3rd May 2004 13:12
All in one solution? eh, well, Waves PAS meters has all, Frequency meter, stereo signal meter, Db meter with Peak/RMS and all the stuff you need, but that's an expensive
Posted: 3rd May 2004 13:24
well here it goes: imo dfx rmsbuddy and inspector are great for quick projects (actually I ues both of them as compensation for lacking features in inspector) - well actually when there was some beta discussion on vowengo plugs, I smartassedly proposed idea of a metering pluggo that would weightedly display freq spectrum (with avg and peak options like in curve eq) and also calculate peak, avg peak, avg rms and peak-to rms ratio. Dr. Alexey agreed that could be a good one (who wouldn't ), but there's no trace of it yet. Oh, yeah, and it was supposed to have a blue led. And be affordable.
Posted: 3rd May 2004 14:20
I'm using Inspector as final insert in the Mastering chain (only have 6 slots, and they're full really fast if I the sound is really muddy). An all in one solution would be great, even if we talk about slotissues. It's not I don't like RMS buddy but if meters were in it and a slightly other layout, I'd use it instead of Inspector - or Inspector instead of RMS Buddy if the stated issues are cleared.
Thanks for the suggestion. Even with Elephant.
Well... PSP Vintage Meter is very good. Nothing to say against it (tested it too). But it's a reproduction of an analog VU Meter. And I'm looking for a digital one. Thanks anyways.
And Waves Plugins will never see any light on my HDD again. I've tested the demo plugins (they wouldn't even properly uninstall!): They're overpriced (not to mention outside of my pricecategory) and color the sound simply too much in my opinion. Sorry.
Well... this looks really promising. If it's that good when it's out and it's free or let's say 20Euro - I'll surely take a look at it.
Posted: 3rd May 2004 14:33
IF it's ever. and IF it has a blue led.
Posted: 3rd May 2004 14:45
My vote would go for Voxengo Curve EQ for frequency analysis and Elephant for RMS peaking too. They are fantastic plugins in their own right and as someone else said, well worth having.
I don't like the Waves PAZ stuff because the graphic display response seems pretty piss poor, very slow on the update (no, not because i have a slow computer, a P4 2.8HT with Radeon 9600Pro graphics card and other graphics meters fly) so I'd avoid that.
For me by far the most accurate and responsive metering has to be the meters in Wavelab 4. They totally rock and it has precisely the kind of realtime RMS meter you're looking for! (with values for peak, rms and rms max and min). Trouble is of course, this isn't a plugin, it's a part of wavelab program itself.
Maybe Steinberg could release those level, phase scope, bit depth etc meters as seperate VST plugs, that could be inserted into other hosts, now there's an idea!.
Posted: 3rd May 2004 15:01
I agree with your assessment that Wavelab 4 has the best metering available.
Posted: 3rd May 2004 15:08
Aren't the Wavelab 4 Metering Tools working as VST Plugins? At least in Wavelab 5 they will be loaded via the slots. At least I've seen so at Musikmesse 2004.
Hm... damn that I still work with Wavelab 3 though. Perhaps it's finally time to look for a higher version. And you can find still some treasures if we talk about "used".
Anyways... perhaps someone else a suggestion? The more the better. =)
Posted: 3rd May 2004 15:14
Seriously?, I didn't know that, that's ace if they have!!. Means I can have them in Cubase too if they've turned them into vst effects.
DO upgrade from Wavelab 3, being able to have all those really useful meters all on screen all the time without being in audio montage mode is a real godsend.
Posted: 4th May 2004 11:39
Well... tested Wavelab4 at my friends place now. Unfortunately the metering tools aren't aviable as VST plugins. They're dll's though, but not VST/DX.
I compared the four earlier mentioned metering tools:
- dfx RMS Buddy
- Elemental Audio Inspector
- vb-audio VU-Meter (Standalone Demo, elseways VST/DX)
- Wavelab 4 VU-Meter
The following snapshots are my results from a song I made around one year ago. I tried to master it in the K-System - so I aimed at K-12. You'll see a shocking difference. Especially with the Wavelab 4 VU-meter. While I eMailed with Mr. Bob Katz from "Digital Domain", he mentioned regarding Wavelab 4:
Unfortunately it seems to be true because every other RMS Meter shows an other result. Even the one from the global analysis. The question now is: which one of those 3 VST Plugins I tested are correct? A word from the programmers would be really great.
Here the two snapshots. Sorry for the long loading times:
Metering Snapshot 01
Metering Snapshot 02
Guess my only choice is really spending 115Euro (without hardware key) for vb-audio. Or waiting (forever?) until Inspector has a new update. hm...
Posted: 4th May 2004 11:49
You are using version 1.1 of RMS Buddy, right? Earlier versions were known to incorrectly measure RMS.
Also, I never look at the continual RMS level in RMS Buddy. The only usable reading, in my opinion, is the average level, and I usually read it while looping the loudest section of the material.
My two cents.
Posted: 4th May 2004 12:16
Hm... an update should do. (I'm still using v1.0)
But still... An average RMS? May I ask for what you need that. It only shows something in between. For the K-System it's useless, IMO. Here I need the maximum RMS level to indicate if my song is in K-20, K-14 or K-12.
Posted: 4th May 2004 12:47
As I understand it, the K-System is a measurement of the difference between average and peak RMS. You need to know where both levels are to know if you're hitting the desired K-System scale.
I believe that in the images above, the red part of the scale indicates the part of the dynamic scale where peaks are acceptable, and the yellow part of the scale indicates the part of the dynamic scale where average RMS should lie. Of course, the diagrams above imply a monitoring system calibrated using hardware VU meters (like the Dorough meters). We're talking about metering plugins, and I'm not sure the same calibration rules apply. Then again, I'm not entirely sure how to apply the K-System to a metering plugin! My assumption has always been that to get a song to meet the level criteria of, for example, the K-14 scale, average RMS should be around -14db FS, with peak RMS sometimes reaching as high as -6db FS. The RMS Buddy and Inspector plugins both deliver the info necessary to make this determination.
If I misunderstand this concept, please correct me -- I want to know!
Posted: 4th May 2004 13:03
The article linked below is the one that gave me the impression that you could easily set up the Inspector plugin for K-System monitoring. I suppose I should ask Bob himself whether or not this guy is full of it!
Posted: 4th May 2004 13:17
Well what I learned (and this one got aprooved from Mr. Katz) - here's a copy from my eMails:
Mr Katz answered this:
So if I want K-14 (CD Ready), my max. RMS peak shouldn't be over -8dB with a RMS meter. Easy math. With the song in my snaps I aimed at K-12 (which would be a maximum RMS peak of about -6dB).
At least this is what I understood. If someone finds a bug in my equation, I'm glad to be corrected.
Regarding the article:
I guess it's time to write Elemental Audio again if this is true that the maximum Headroom is +4dB. Like I said... I used Wavelab as basement for my metering and if I use this equation mentioned earlier, I should get this result:
+6dB - K-14 = -8dB max RMS
But if I trust the article, then it's:
+4dB - K-14 = -10dB max RMS
Now... either someone is totally confusing us, or I was totally wrong from the beginning. If you can ask Mr. Katz to join our discussion, I'd really apreciate it, Lazlo.
Posted: 4th May 2004 13:22
FWIW, I just emailed that article to Bob Katz asking him if he feels that is a proper way to determine whether or not levels are meeting the K-System guidelines. I hope he answers.
Posted: 4th May 2004 18:14
Thanks for informing him, Lazlo. However there's something very fishy in the article you posted earlier. Let me quote what I mean:
This can't be happening. If Inspector only has a 4dB headroom, the whole K-System equation would be turned upside down and 2dB off. Afaik standardised VU/RMS meters have a +6dB headroom (PMM). Same counts to 90% of all aviable sequencers and audio editors. Inspector's PMM just goes up to +0dB, but this isn't an issue because the "host PMM" has more priority and shows the complete headroom anyways. This "mistake" in the article can really confuse the reader.
But back to the topic. I took some time testing the earlier mentioned plugins again. With the main idea in mind "I have to test my equations for the K-System again if I'm wrong or not".
For this testrow I used some "pop songs" run through our "usual suspects":
- dfx RMS Buddy
- Elemental Audio Inspector
- vb-audio VU-Meter (Demo, but only limited for running length)
- Wavelab 4 VU-Meter (which is 3dB off)
- Wavelab 4 Global Analysis
The equation again:
PMM - K-System = max RMS peak
I only looked for the RMS levels and categorised the tracks itself into the IMO fitting K-System after the metering. Please also keep an eye on the vb-audio VU-Meter results. They were the most precise ones IMO. Second best plugin is "Inspector". One thing is for sure... even though my testfiles were extracted from MP3 files (192kbit and upwards) and not from CD, they did their job. But you'll also see massive differences if we call about mastering.
Here are the results:
Kumi Koda - 1000 no kotoba (Original Mix, FF X-2 Soundtrack)
dfx RMS Buddy: -11,00db (average)
Inspector: - 6,00dB (max peak)
vb-audio VU: - 6,00db (max peak)
WL4 VU: - 9,46db (max peak)
WL4 analysis: - 5,78db (max peak)
Camouflage - The Great Commandmend v2 (Single)
dfx RMS Buddy: -12,38db (average)
Inspector: - 7,50dB (max peak)
vb-audio VU: - 8,06db (max peak)
WL4 VU: -11,01db (max peak)
WL4 analysis: - 6,65db (max peak)
Die Fantastischen Vier - Le Smou (Video Version)
dfx RMS Buddy: -10,70db (average)
Inspector: - 5,30dB (max peak)
vb-audio VU: - 5,08db (max peak)
WL4 VU: - 8,83db (max peak)
WL4 analysis: - 4,83db (max peak)
K-SYSTEM: K-12 (if it would exist: K-10)
Limp Bizkit - Eat you Alive (CD Version)
dfx RMS Buddy: - 9,54db (average)
Inspector: - 5,70dB (max peak)
vb-audio VU: - 5,71db (max peak)
WL4 VU: - 8,01db (max peak)
WL4 analysis: - 5,31db (max peak)
Justin Timberlake - Rock your Body
dfx RMS Buddy: -12,61db (average)
Inspector: - 5,80dB (max peak)
vb-audio VU: - 6,43db (max peak)
WL4 VU: -10,34db (max peak)
WL4 analysis: - 5,17db (max peak)
Alison Krauss - Down to the River to Pray (O brother where are thou OST)
dfx RMS Buddy: -19,24db (average)
Inspector: -10,00dB (max peak)
vb-audio VU: -10,33dB (max peak)
WL4 VU: -14,58db (max peak)
WL4 analysis: -10,81db (max peak)
K-SYSTEM: K-14 (is between K-20 and K-14, so the next higher system is valid)
Scooter - Level One (Techno Stadium Experience Album)
dfx RMS Buddy: -10,10db (average)
Inspector: - 5,85dB (max peak)
vb-audio VU: - 5,62dB (max peak)
WL4 VU: - 8,14db (max peak)
WL4 analysis: - 4,75db (max peak)
Unfortunately I couldn't use dfx RMS Buddy v1.1 for this test. I downloaded the actual version from the DestroyFX page (which is dated November 2003) but the version in the zip is still the old v1.0 from February 2003.
Posted: 4th May 2004 19:30
Very interesting, compyfox. Thanks for these tests!
I heard back from Bob about the Web page that describes using Inspector to monitor RMS.
It seems that Bob doesn't think a piece of software is enough -- your monitoring must be calibrated in his view in order for you to be able to make precise level adjustments according to his systems.
That makes sense -- two pieces measuring the same average RMS level might not have the same apparent level. One might have more mids or highs than the other, leading the listener to perceive it as louder than the other.
Posted: 5th May 2004 05:39
Hm... this is true. To a good "metering system" also counts a good monitoring. It is true that every system sounds "different" if it isn't calibrated. That's why I nailed the issue down to just our little plugins. So the metering is very linear because everything is "inside one machine" (if you work purely digital, or if this is your final mastering).
In the tests I made, the two plugins "vb-audio VU-Meter" and "Elemental Audio Inspector" especially shined though because of their RMS meter. From the free/low budget ones they're really the best aviable ones in my opinion. And they work perfectly for the K-System.
BTW: I wrote down the K-System at the end of each tested song before I ran everything through the vb-audio VU-Meter (because this was the demo that only runs for 5 minutes and only 1 minute per tested song). I not only used it for very precise metering, I also used it to see if I was wrong or not. So my math isn't off. Like I said... there is a bug in the article.
Posted: 5th May 2004 07:26
Compyfox, I think the article is correct except for one thing, Inspector measures RMS without a +3dB correction that is sometimes made. This correction is made so that a full scale sine wave reads at 0dBFS rather than -3dBFS. The correction is optional, but it is needed if you want your meters to conform to the K-system. This means that in Inspector the rms "caution level" needs to be set at -17dBFS and the "warning level" at -13dBFS to correspond to K14.
Posted: 5th May 2004 07:33
Wow -- great info, Andrew. Thanks!
Posted: 5th May 2004 07:56
But one thing eludes me, Andrew. If this is true... then why does Inspector show the same results than vb-audio VU-Meter? And I used Inspector with Wavelab 3 and 4, the VU-Meter demo was standalone and showed the same.
I didn't set the caution levels. I only looked what K-System I'm aimed at, and scrolled the warning indicator from Inspector to the maximum allowed peak (for K-12 it's -6dB for example).
Did I get something totally wrong here?
Posted: 5th May 2004 08:16
They are just two different ways of measuring RMS -- you either apply the 3dB "correction" or you don't -- it's optional. But the k system specs are referenced to rms meters with the 3dB correction included.
A way to check this is to play a full scale sine wave and see what the rms meter says -- so if it says -3dBFS it hasn't had the correction, if it says 0dBFS it has.
The basic concept (as far as metering is concerned) in the k-system is that for K14 -14dBFS is the "zero level" and you have a 4dB "headroom" above that so that the loudest RMS peak you should see is no higher than -10dBFS. If the RMS meters do not have the 3dB correction then those values become -17dBFS and -13dBFS. Hope that is understandable.
Posted: 5th May 2004 08:58
So, to clarify, no that's not right -- for two reasons. The k-system defines a headroom of 4dB not 6dB, so for k-12 the warning should occur at -12+4=-8dBFS. But also because of the way Inspector's meters are calibrated (no 3dB "correction applied) the warning level needs to kick in 3dB lower, so -12+4-3=-11dBFS
Posted: 5th May 2004 09:10
Now you're totally confusing me. I always used Wavelab for metering which has a +6dB headroom. I did the math, metered with both plugins and both showed the same results. As you can see with the results with the tested pop songs.
Can you please download the two metering tools, test them and post some snapshots? Perhaps even the WAV file you used for testing for reconstructing? I hope I get what's wrong afterwards (am I shuffling something with dB, dBFS and dBU?). It was so clear the last time I talked with Mr. Katz.
Posted: 5th May 2004 09:20
The headroom is defined by the metering scale, not the meter itself. By that I mean it is an arbitrary figure which is used to interpret the meter's results, it is not integral to the meter itself. I'm not familiar with the Wavelab meter, but I have tried the Inspector -- in Inspector when the meter indicator goes to the top of the scale that means the sound is at 0dBFS, now we take the k-system scale and apply this to the meter -- the k-14 scale says that rms at -14dBFS is equivalent to a zero value and that rms at 4dB above that (-10dBFS) is the warning level. So the meter should be coloured so -14dBFS to -10dBFS is amber, and -10dBFS to 0dBFS is red (it should also show -14dBFS as a 0dB level, but that is not possible in inspector).
Edit: I'm ignoring here the fact that Inspectors RMS will show a value 3dB below the k-system specs...
Have I confused you more, I hope not?
Posted: 5th May 2004 09:35
I've thought of a different way to explain this -- from the beginning :
Lets make an rms meter, the top mark on this meter is 0dFS, so if the meter is lit all the way up to the top, the sound is at 0dBFS. When we calculate the RMS value we will add 3dB to it (so that it confirms to k-system specs). This means that a full scale sine wave will light the meter up all the way to the top.
OK we have our meter, now we need to calibrate that meter according to the k-system, so for k-14 we put a value of 0dB at the -14dBFS point of the scale. The "led's" on the scale below this point are green. We now add a point which is 4dB above and colour the led's between 0 and 4dB (which is -14dBFS and -10dBFS) amber. All led's above 4dB (-10dBFS) are coloured red.
We now calibrate our listening environment so that pink noise at a zero level in our snazzy new meter is 83dBC SPL.
From a metering point of view the meter should not go into the red, but should hover at or below the zero level, peaking into the amber.
Does this explain?
Posted: 5th May 2004 11:04
Okay... I guess it's clear now (after some handdrawings) how to set up my warning ranges and how a K-System based RMS meter is build up. But still there is the question that confuses me:
Why do I get the same results with my math in Wavelab with Inspector and with the standalone VU-Meter Demo from vb-audio.
I made two new snapshots just with Inspector, the VU-Meter from Wavelab and my master PMM with the inserts. Both snaps use the same song: Justin Timberlake - Rock your Body. The first snapshot shows the song without anything altered. Just the raw plugins and mesuring tools. The second snap shows the measuring tools and (what's not on the picture) an activated compressor that boosts the sound +4dB upwards (for loudness - usually you cut it up at +0dB, but I just want to show the effect).
As you can see in this example, even though the PMM in inspector only goes up to 0dB, it still has a +6dB headroom like Wavelab too. Okay... so much about the set rules.
Now back to the math:
I have a PMM that reaches from -96dB up to +6dB. Which gives me a +6dB headroom, that should resemble 0dBU if the peak reaches fully up.
I want to use K-14 (Pop in general, CD-Ready). Which means I have to subtract 14dB from my full range. This gives me the following math again:
Okay... so now I have the maximum peak for the K-System and my caution level starts with -8dB (amber).
Then I have a 4dB headroom "for" this system as warning. Meaning:
So my warning indicator (red) starts at -4dB.
For K-12 it's:
For K-14 it's:
For K-20 it's:
This is at least what I think that is right (and Mr. Katz agreed with the words mentioned a couple of threads earlier). Well... I don't have Pinguin Audio Meter, but I should get the same K-System results, shouldn't I?
Corrections apreciated of course.
*uhm... if so, then please as simple as possible if you don't mind - I'm sure not everyone understands our "chinese-talkie" here (not to mention that I'm slowly getting confused too... erm...)*
Posted: 5th May 2004 11:33
I don't know all these other meters and I don't know what these +6dB marks are measured in relation to.
The only way that you can be sure of the meaning of the scales on these meters is to play a full scale sine wave through them. If the peak reading and rms are the same then the meter is adding the 3dB correction to the rms (which is good for the k-system).
But, whether or not the peak and rms correspond, whatever value the rms meter shows for a full scale sine wave, that is the value which corresponds to the reference below which you place the k-system scale, so the k-12 will have a 0dB point 12dB below below the value shown by your rms meter when monitoring a full-scale sine wave.
If the rms has had the 3dB correction then you'll have a well functioning k-system meter, if it hasn't then although the rms part will show the correct amber "safe zone" etc. it will not be being displayed at a correct level relative to the peak meter.
If you use this method then you won't have to worry about the vagaries of different manufacturers meter labelling.
You can generate a full scale sine wave with Voxengo's deconvolver:
Post pics of the full scale sine wave -- hopefully that'll make things crystal clear
Posted: 5th May 2004 14:35
Okay... now we have some snaps for reference. I tested Inspector, Wavelab 4 VU Meter and the vb-audio VU-Meter. All three are 3dB off. PSP Vintage Meter didn't give a good result at all btw. and the Wavelab global analysis (no matter if WL3 or WL4) also show 3dB off.
Here is the mono sweep I used for testing (generated with Voxengo Deconvolver):
Sine Sweep (44.1kHz, 16bit, 30s)
If someone could test this with Pinguin and give us the result from the PMM and RMS, this would help a lot. Thanks in advance
So... setting up the meters for correct K-System metering shouldn't be a problem anymore if we relate to the information you just gave us, Andrew:
However... if there are my settings now, all the meterings I made are wrong. For example "Mr. Timberlake" and "Mrs. Krauss" from page 2 again.
If I use my math, then the K-System is correct. Because if we look at the image from DigitalDomain again:
The highest values resemble in my opinion to +0dBU (which is the maximum peak from the complete aviable PMM scale including the possible headroom, in Wavelab for example -96dB to +6dB). Now if we do the "PMM - K-System = max RMS peak" thingy again, we get the results like in the second quote from our popstars.
But if I use the new settings based upon the last informations you gave me and compared the upper list too, nothing is correct anymore.
Now is the question:
What else is correct?
What are the right maths?
I mean... if it's that way, then all mastering studios that engineered these songs are totally wrong, aren't they?
*I guess we can really need Mr Katz at this board now, as he made the K-System*
Posted: 5th May 2004 14:47
Well... please do keep in mind that the K-System is rapidly being adopted, but currently very few mastering facilities are following it to the letter. Do you know for a fact that either of these records were mastered with strict adherence to the K-System? My guess is that they were not.
Posted: 5th May 2004 15:16
Well... don't have a definite proof. But I'm sure Kumi Koda (Virgin Japan and DigiCube), and the studios where Timberlake, Bizkit and Mrs Krauss are, using the K-System. Soundtrack firms are known for working with set standards. And because Kumi Koda made tracks for "Final Fantasy X-2" and Alison Krauss for the "O brother where are thou" OST, I'm sure they're quite in the standards.
Besides. Can you please ask Mr. Katz again to join our conversation or at least watch over this thread here and giving some suggestions, Lazlo? I guess this would clear things up a lot - not only for me.
Posted: 5th May 2004 15:57
You need to click on the PSP Vintage Meter name at the bottom of the plugin -- this opens a configuration page -- you will see that the zero scale is already offset -- if you offset the reference to -14dBFS (and set the integration time to 600ms) you'll get "sort-of" K-14 type meters. If you set the reference 0dBFS then a full scale sine does read at 0dB.
These settings are correct.
WRT to whether or not the music you tested was mastered according to k-system, I doubt it very much. I can't remember where I read it, maybe on Mr Katz' site, but there are examples of film soundtracks where if you play the DVD at a sensible volume, and then try the CD album you'll get blasted out of your seat -- in other words just because the standards are followed for the sound that accompanies video, it doesn't mean that the same piece of music will follow the same standard when it's audio only.
Using your measurements (from Inspector), and K-14 scaling: the Timberlake track RMS peaks at +11.2dB (very hot), and the Krauss at +7dB (better, but still HOT). I don't think there's anything unusual about these sorts of figures with modern cd's. After all, one of the purposes of the K-system is to provide a common standard to end this pointless volume war.
Posted: 5th May 2004 16:08
I think this is where the confusion lies, because there is no inbuilt headroom above 0dBFS -- go above that and you get clipping. What I don't understand is why some of the meters you have posted pictures of have scales which appear to be in dBFS (which I think is proven by the sine wave test) but which have a red area which goes up to +6dB. In my mind this is totally bogus (unless these meters are capable of oversampling -- which I don't think is possible for a host based meter). If however there is a setting in these meters which allow you to offset the 0dB value to a level other than 0dBFS, then having a red area does make sense -- but I am not familiar with these meters, so I don't kknow if this can be done.
Listen to a classical cd and see if the k-system scaling as I've explained it makes more sense. Don't judge what I'm telling you on overcompressed Timberlake, there is no way that that sort of stuff is mastered to k-system standards.
Posted: 5th May 2004 17:35
Here's a quote from the english helpfile of Wavelab:
I usually see this as my "loudness headroom" or for "soft clipping". In Cubase for example it's often very helpful for still having the option for making and instrument a tad louder if needed, thanks to the +6dB Headroom. It really only hardclips when I get over this level. Elseways it doesn't distort at all.
Besides: Wavelab 3 and Cubase 5.x had a dB range from -76dB to +6dB (82dB max), and Wavelab 4 has a dB range from -96dB to +6dB (102dB max). And they're standardised production and mastering tools.
Wait a minute. How do you get these results?
Anyways I thought about the "warning" indicators. Could it be that the 4dB headroom also used in most cases to keep the stuff "as hot as possible"?
You know Murphy's Law, right? So please take a look at that following snapshot with the song from Camouflage:
Resemble the Winamp VU with the Inspector. I only set the "warning indicators" and didn't chance anything else. But there's something very fishy about it: even though I didn't catch it in the snap, Inspector's RMS went into the range of -8dB. With the "3dB off" list it's way off the track. But Wavelab VU showed me -11.34dB. From Mr. Katz I know for sure that the Wavelab 4 VU meter is 3dB off. But Inspector reacts on the "correct meter", not the "3dB off" one
Er... so actually what's happening here?
Regarding the classic songs I tested two other tracks. Here are the results. As you can see, the measuring tools greatly differ again.
Johann Pachebel - Canon C Dur (Evangelion Soundtrack, Remastered)
WL4 PMM: - 0,20dB
WL4 VU: -12,56dB
WL4 Analysis: -10,33dB
Just because of the volume level (which reaches 0dB) I'd count this to K-14. But k-12?!
Johann Sebastian Bach - Menuett in G-Dur
WL4 PMM: - 1,80dB
WL4 VU: -11,87dB
Inspector: - 8,40dB
WL4 Analysis: - 9,96dB
I'd count this track to K-20, max K-14 (because it get's over the K-20 warning level if I go by the "correct metering" list) but with this 3dB off measuring it would count to K-12?!
I mean... if every track I have on CD is K-12, especially classic that usually never ever reach +0dB PMM (which is an indication that it's between K-20 and K-14), there's something really fishy about it, don't you think? Same counts to "Mrs. Krauss" btw, the track doesn't peak higher than -1dB PMM.
If this case is true, even tracks from me or my friends that were rendered with -3dB in Cubase (for better mastering) would show that they're already K-14? hm...
Sorry for my rudeness but may I ask how long you already work with the K-System or how far you're influenced with it? I mean... after having longer conversations with Mr. Katz this thread totally throws away what I thought was right and also was confirmed. So please don't take this personally.
Nothing against the "loudness war", but this simply can't be happening if half of my tested tracks who're only half a year old at max and from CD while the K-System is already a couple of years old. And still they're all off track?
Perhaps there're any recource files existing that're already in a specific K-System? This would also help a lot IMO.
Posted: 5th May 2004 18:09
So do the clip lights come on when the meter goes above 0dBFS or above +6dBFS?
Remember, there is no such thing as "soft clipping" when you clip a digital signal -- i.e. when you bounce down a 32bit float file which has gone above 0dBFS.
For Timberlake: you say inspector RMS peaks at -5.8, now you have tesrted a full scale sine wave and it shows as -3dBFS in Inspector so for k-14 the zero level is -14-3=-17dBFS. -5.8--17=11.2
In the end I'm not going to convince you that I'm right -- I understand your dilemma -- either you believe what I'm telling you and you have to accept that most cd's are over compressed (according to the k-system) or don't believe me and be happy that most cd's do fit into the k-system just fine.
I use RME Multiface which has k-system metering built in -- if you let me download a piece of music I'd be happy to test it and send my results to compare to yours with my interpretation and yours.
I don't think there's much more I can add, perhaps someone else can chime in.
Posted: 5th May 2004 18:26
above 0dB Full Scale
It turns into a square wave, true. I'm just used to call it that.
Ah... okay, this is understandable. I'm only used to negative scales because of Inspector and such.
Oh please don't take this personally. I'm just generally stubborn and try to find my way through the jungle by asking from both sides. I hope I didn't threat you.
One of my own songs or one of the testsongs I used? Thanks for the offering, I really apreciate that.
Well... you can add something. I don't know if this is possible, but do you have or can you produce reference files (for example actual songs, soundtracks or just demosamples) in the K-System for testing and upload it somewhere? For analysing and stuff. I guess a lot more readers are also interested.
Thanks in advance... and I totally forgot to thank for all your suggestions and opinions.
Posted: 6th May 2004 02:50
Ok that supports my supposition -- the 0dBFS to +6dBFS as used in those meters are indicating the level of the 32bit float file -- with a 32bit float you can exceed 0dBFS with no distortion (the headroom above 0dBFS is enormous), BUT as soon as you bounce it down to, or pass it to your soundcard as, 24 or 16bit integer (which you have to do to), then anything above 0dBFS will clip. In other words there is NO headroom above 0dBFS.
No problem it was getting rather late
I don't mind, just so long as we're using the same audio file.
Posted: 6th May 2004 13:29
Well... as long as I produce in 32bit float and render it down to 24bit or even 16bit with a diphtering matrix, I should be on the save side. But this is a completely other topic.
Hm... do you mean the RME DIGICheck by chance? Wasn't the K-System support just implemented after Musikmesse 2004 (or while). Looks promising. However I don't have a RME card but Terratec. I'd love to see a tool like that for the Professional Series too.
To the testfiles:
I uploaded a pack of different files for you to test. They're all from a remixing community I'm in that revives old videogame music with a new coat. But that doesn't mean that we in our community don't have the same problems with "volume war", proper mixing and stuff. However... you can choose if you want to use a WAV or a MP3 (320kbit) from this samples for testing. I also made snapshots from my meterings again for resembling. I go a bit more in depth after the filelist:
DOOM - Demonik Elektronik (OC Remix)
WAV Version (ca. 12,80MB)
MP3 Version (ca. 12,80MB)
This track was produced with hardware and software. Afterwards mastered in Logic with Waves plugins by the producer himself. The track is very hot, but fortunately it's peaks didn't transform into a "squarewave" - even though the track looks like it.
Final Fantasy 7 - Sector7 Hath wrought the Angel (OC Remix)
WAV Version (ca. 17,35MB)
MP3 Version (ca. 2,67MB)
This track was produced completely digital (Fruity Loops, SoundFonts, Edirol HQ Orchestra). I mastered it with Wavelab 3 and tried to put it into K-14 (with my math: +6dB - K14 = -8dB RMS max). It's still very dynamic, though the end sounds like it isn't. But this was a fault from the producer. He likes to use massive compression and I had a lot to do to "unmud the sound" again. I simply had to force him to send me a 50% more quiet song for proper afterediting - so he just lowered the volume level and rendered it down to WAV. For mastering I used a 30band EQ and a compressor just to raise the loudness just a tad.
SaGa - Let the Saga begin (OC Remix, Beta 4, unmastered)
WAV Version (ca. 12,84MB)
MP3 Version (ca. 2,25MB)
This track was produced with hardware and software. Recorded at 0dBFS with an EDIROL recording device. No changes afterwards. You can say this is the RAW version.
SaGa - Let the Saga begin (OC Remix, Beta 4, mastered)
WAV Version (ca. 12,80MB)
MP3 Version (ca. 2,35MB)
This is the "mastered" version. I tried to put it into K-12 on purpose (with my math again: +6dB - K12 = -6dB RMS max). It's hot but still very dynamic. And even though my math is "wrong" I'm still in K-12 (if my meter is correct, see theory further down), I just used up the complete headroom (amber zone). The "natural loudness" from the strings let it go a bit wild, but I think I could handle it. For mastering I used Wavelab 3, just a 30band EQ (to clean up the sound a bit) and just a compressor for raising the loudness a bit. I didn't want to destroy the feeling.
All meterings at that time were made with Inspector only btw. Not the FFT, but the RMS Levels.
Which brings me to the next point. Please don't think that I refuse to trust your words, Andrew. But this really eludes me. I tested Wavelab 4 VU, vb-audio VU-Meter and Inspector all with a generated 0dBFS sinesweep. All showed the same results: 0dBFS by -3dB RMS. While I was on the RME page to read a bit into DIGICheck I also found testfiles for looking if the meter works correctly. I downloaded the wav samples 0_16 and crest_16. Again, all meterings tools showed the same (0dBFS by -3dB RMS, with the crest one 0dBFS by -19,2dB RMS - which seems to be 3dB off, unfortunately ).
Here's a quote from the RME page with the testfiles.
So the RME meter should show the same, shouldn't it? But regarding the K-System the meters are calibrated with a +3dB correction so that both the PMM and RMS show +0dB.
So far so good. But there's really one thing strange: Even though all tested meters show the same with a sinewave test, while metering a song: Inspector and vb-audio showing 3dB more than Wavelab VU. Even Wavelab global analysis shows 3dB more. Just compare the pictures from my meterings. It's really kinda scary but they really do.
Mr Katz also asked me to post this after I sent him an eMail last night:
BTW: He tries join our conversation when he gets a free minute. He's actually stuffed with a lot of work.
But back on topic. I don't know why this is so, but this would mean that the article posted from Lazlo is correct: K-System for the digital studio, written by Niko Nyman
And I can use the values without the -3dB correction.
So all I can do now is ask to test the files please and post me your results for comparing. If your values with the K-System meters show (almost) the same as the ones from vb-audio VU-Meter and Inspector, then the plugins work correctly and no "correction" should be needed.
Or we all were totally wrong from the beginning.
*if someone has Pinguin Audio Meter, I'd love to hear the results too. Thanks*
Posted: 6th May 2004 14:01
Actually, I would NOT assume any of those records were mastered to any system other than the "oh SHIT, this is gonna be on the radio so let's peak limit the FUCK out of it" system. The fact that Mr. Katz has proposed the K-System is exactly because of top 40 records the likes of those by Mr. Timberlake and Misters Bizkit. I think K-System adoption is only really catching on in the wake of Mr. Katz's book being published, and I seriously doubt most of the major label engineering houses will pay much attention to it, at least initially. "Louder is better" is going to die a long, slow death, in part because major label executives understand it -- they're concerned with initial impact, not overall quality. It is independent musicians and audiophile engineers who will be giving credence to the K-System in the early adoption stages (which I believe we are still in).
I could ask him, sure, but it's not like he's a friend of mine. He responded to your emails, as well, so I think you have as much of a chance of convincing him as I do. Still, I'll try to email him today.
Posted: 6th May 2004 15:34
Dithering will NOT stop a 32 bit float file which has a level greater than 0dbBFS from clipping!!!
Compyfox, I don't have the time to respond to the rest of your post right now -- hopefully tomorrow (Friday) evening I'll have some stuff to post for you.
Posted: 6th May 2004 16:01
That's okay. Thanks for the effort though.
If we can pin down the issues with "Inspector" and "vb-audio VU-Meter", perhaps we can even help the plugin developers improoving their tools.
Looking forward to the results.
Posted: 7th May 2004 13:48
Compyfox, I'm not going to use any of the files that you gave links for, so sorry for wasting your time there. The reason is because it's virtually impossible to capture a meaningful moment in a real piece of music, due to the way that different meters respond to musical transients (meter ballistics) -- more on that at the end.
Instead I have posted a picture of a sine wave at -9.0dBFS. The Digicheck meter is set to K-14, PSP VU meter is set to a -14dBFS offset, Inspector is set to show amber at -17dBFS and red at -13dBFS.
You will see that all three meters correspond exactly, so as far as I'm concerned this proves what I have said above.
So why are you finding your different rms meters peaking at different levels even when they are aligned when measuring a sine wave? Assuming that the meters are functioning correctly, the reason for this is meter ballistics, a meter will be set up to move faster or slower -- typically an analogue VU meter has a fairly slow ballistic, so it can't respond to very short peaks. Ideally a K-system meter will have a meter characteristic which reflects the human perception of loudness -- Bob Katz suggests an RMS meter with an integration time of 600ms (you can set the integration time of PSP's meter) which is even slower than a standard VU, but it probably more accurately represents the loudness that we, as humans, hear. This means that comparing meters by instantaneous rms peak levels is probably not very helpful, because this figure is very dependent on the meters' ballistics and these are very variable (to me it looks like the ballistic on Inspector's RMS meters is fast compared to a standard VU, and therefore peaks higher). If you try PSP's meter and fiddle with the integration time parameter you will see exactly what I mean.
Posted: 7th May 2004 14:30
Unfortunately this time I'm stuffed with work. I'll try to answer saturday evening/night. I hope this is okay.
Posted: 7th May 2004 17:32
If that's music
Posted: 8th May 2004 04:54
More thoughts: the ballistics on Inspector's RMS meters are so fast (they are also very thin and hard to see) that I don't think they are very useful for k-system type metering. A much better option is to use PSP's Vintage Meter -- set integration time to 600ms and set 0VU to reference -12, -14 or -20dBFS, and click on the meter face so it shows the full -40 to +6dB range. With these settings the meters give very similar readings to RME Digicheck (with Digicheck's ballistics set to RME's recommended 20dB per second). If you do this then PSP's 0-4dB range corresponds to the k-system amber zone, and anything above +4 corresponds to the k-system red zone.
Posted: 10th May 2004 10:34
I never would've guessed that VintageMeter would be the ideal K-System meter, but I'm sold. I've set it up as you described and am really happy with it -- very easy to see where average levels are!
Thanks for this invaluable info.
Posted: 10th May 2004 12:32
Lazlo, yeah it's a great little meter and free too!!!!
You'll now have to put up with everyone telling you that your cd's are too quiet
Posted: 10th May 2004 12:57
Ha! Too true. Well, what I intend to do is put up an explanation of the K-System on my mastering Web site and explain that though I will increase level if the band demands it, there are many more good reasons not to do that. Hopefully that will sway a few people. But I'm not too hopeful. Ultimately, the customer is always right... damnit!
Posted: 10th May 2004 13:00
It's too bad I've never got VintageMeter to work properly
Posted: 11th May 2004 01:05
just get a look on the debat here.
let say one or two things.
to test an RMS / PEAK meter , it's simple : generate a sinus 100hz (or whatever frequency) with a gain of 1.0 (=0db peak) . then the RMS meter will display -3.0db (otherwise it's not a RMS meter)
wavelab will display this -3db too, but needs several seconds of sinus sound to reach it (between 2 to 4 seconds).
VB-Audio VU-Meter is giving you the instantaneous RMS level, the others try to merge VU-concept and RMS concept by adding attack time and ballistic stuff... this can be interesting, but this is depending on nothing in term of norm or standart...
Posted: 12th May 2004 05:11
Askywhale DetectComp Analyser/Monitor is all you need show Realtime multi-criterion numerical analysis :
High and low velocity.
High and low frequency mean.
Compression statistical indicator.
Peak and RMS progressive values.
Velocity Distribution computation.
Frequency Distribution computation.
Velocity and frequency evolution.
Posted: 21st June 2004 14:44
I just had to dig out this thread for some further suggestions, theories and of course... questions. First of all I have to excuse myself for my lack of reading and answering in here. Murphy's Law.
Anyways... I found the time today to get back into the topic and test out some stuff. To be honest: Akywhale's DetectComp isn't suitable for metering with the K-System in my opinion. Perhaps it's good for overall view. But the FFT and all other spectrums react zero to nothing. Besides... it's also 3dB "off", which isn't suitable for our purpose.
Then thanks for the answer, Vincent Burel. However is a K-System implementation planned in VB-Audio VU Meter? If so, I'm considering using it with pleasure.
But back to the topic and: PSP Vintage Meter. I'm getting more and more addicted to it's capabilities, if I may say so. I was amazed of the nearly prezise levels it gave me while testing around 30 tracks today (from mid 80ies track to tracks from nowadays). Okay... it is a rebuilt of a vintage meter, not a digital one, but it is good indeed.
In my todays testrow I was very impressed by several songs. Under them:
- Phil Collins - You'll be in my Heart (Disney's Tarzan OST, very dynamic, K-14 by -3,5dB peak!!!)
- Meat Loaf - I'd lie for you and that's the truth
- J.S. Bach - Brandenburgisches Konzert Nr. 3 G-dur BWV 1048
- Black Sabbath - Paranoid
Most impressive was the track from Phil Collins. With the settings mentioned by Andrew (and set to K-14) it barely reached over +0dB in PSP VU at all. The Track from Black Sabbath and the recording from the "Brandenburgisches Konzert" however sometimes went over +0dB but not over +2dB to +3dB from the set "amber zone" (more to it later).
The track from Meat Loaf (was from mid/end 90ies) was also impressive if I may say so... Loud, dynamic and still in K-12. Well if we see the stated +4dB amber zone as "headroom" for fortissimo (very loud) sections in the song caused through massive layering of instruments, bass/vocal boosts and effect explosions.
While testing actual tracks from the Charts (for example Nelly Furtado, Scooter, Nightwish, The Rasmus) the needle surely exeeded +3dB. But setting the 0VU to -8dB or -6dB (which would be surely overcompressed) it showed me that most of the nowaday mastering engineers at least have a "line" where to stop. But unfortunately for the price of the dynamic.
And here are my points that still elude me to understand the K-System (which I hope that you can help me):
1) AVG - Average Level or Peak?
The question is simple... When I setup PSP VU Meter's backside to 600ms reaction time by -14dB FS = 0VU... what exactly does that mean in general?
Okay... the reaction time is understandable. But does the new set +0dB point mean that the maximum peak shouldn't get over +0dB = -14dB FS or is it allowed to hover around it (which means average level). This is a thing I still need to get to know better.
Then what about the "amber zone" (+0dB to +4dB) and the "red zone" (+4dB and upwards)? Well... if I had a meter like the RME one or Pinguin, this would confuse me even more. But from the tutorial on Bob Katz page it's written that the "red zone" is used for "loud passages like explosions and effects in movies".
Then back to the root question... where should I aim my metering? Around +0dB or +4dB (max "amber zone")? I've heard songs that should have K-12 (like the one from Meat Loaf) but are for a long time in the amber zone or is that only allowed for a handful of seconds so that the track's "peak" should count to K-8 which doesn't exist?
It's a bit confusing, but a serious and important question.
2) Switching settings in PSP VU Meter from VU to PMM
When I switch from my set up VU settings over to PMM, the meter shows me something similar like Wavelab 4 and Inspector. However when I have (let's say) set up K-14 in the VU and the track is truely K-14 (like the one from Phil Collins), then the needle hovers around -10dB to -8dB. A bit confusing, but then again, when you go one or two sides back and recall my math, then it should be correct, like Mr. Katz also said "if the Meter is correct". Or am I on the wrong path again?
PMM is good for seeing if I get clippings and such. However... can I still get correct informations regarding the K-System when I use the VU meter only? Personally I think this should be possible.
Again... thanks for reading and answering. I hope this topic is as interesting as it was on the first day. Not only for semi pro and pro musicians.