Syntrillium Backup Forums :: View topic

kylen · Posts: 290

I was over at the UA web site reading about the 1176LN again Tongue

and decided to give all of my compressors the attack test per the test they have on their web page:

http://www.uaudio.com/emulation/index.html

Anyone else done this? Bear in mind my test wasn't scientific or anything but it didn't take long for Ozone to shoot every compressor I have by taming any possible transient in the audio spectrum with its' very fast attack.

Figuring that an audio spectrum of 10Hz to 22.5KHz includes waves with periods of 100ms thru .04 ms you need a really sharp attack time if you want to be able to catch all transients. So to get a 22.5KHz transient you'd need a 40 us attack time to catch the very first one.

That's in the range of the 1176LN as its' attack can go down to 20 us.

I generated a 20KHz wave with CEP that was similiar (?) to the one in the 1176 vs Bombfactory shootout. According to what I saw Ozones' attack is very impeccable and precise. I used a 0ms attack and it performed its gain reduction and release with no artifacts or overshoots like my other compressors did.

I Use CEP, Ozone, PSP, TRacks, Timeworks, RNC, db-audioware, Voxengo, and digitalfishphines compressors and limiters. They all have their uses and individual sounds.

What I'm interested in this weekend is studying control of transients so that's why I wanted to see if I have anything that will catch them all without producing potential distortion via overshoot. It looks like Ozone does that.

So I'm just basically sharing this with anyone who is interested and wants to discuss these kinds of things. Or if anyone has tried this and gotten other results or successes with other similiarly priced limiters. Or my approach is just wrong - I'd like to hear that also.

The goal here isn't to kill the life-breathing dynamic transients, just to understand them a bit and try to control them within a comfortable range of db or headroom. What I'm finding is that the pop/rock reference material I like has an average rms of about -12db, max rms of -6db, and peaks to 0db using CEP stats w/50ms window (a 50ms window which allows the entire audio spectrum 20Hz-20KHz in).

So I'm looking and listening to what makes up the max rms to peak db range where those type of transients live and how to better apply that to my own material without squishing it to death.

Thanks for any comments of feedback. Pro or Con...:)

kylen

DeluXMan · Location: Canada Posts: 330

It might be worth pointing out here that the CEP compresser, and generally digital compressers have the ability to 'look ahead' and respond to a peek at the time of the peek or just before, allowing the compresser to catch all of the peek if desired. This also allows you to set the attack time to something greater than 0, and still catch those peeks.

post78 · Location: USA Posts: 2887

kylen · Posts: 290

post78 · Location: USA Posts: 2887

Hi kylen. No rolling eyes here! :)

kylen · Posts: 290

OK thanks Post78 whenever you're ready...

I've been over at Digital Domain tonight ...
http://www.digido.com/audiolevels.html

...reading after spending the day trying to balance multi-band rms levels in a full mix to some reference material instead of trying to do it using compression and limiting right off the bat. Trying to recap the days' activities in the world of level balancing while preserving dynamics - maybe there's hope yet !:)

kylen

post78 · Location: USA Posts: 2887

Okay, here are five files, each a small (148 KB) MP3. This was a simple beat that I recorded to try out the new microphone. The set is small and old, just like the room, so don't expect anything great. I set the compressor to careless levels to make the effects more apparent. Anyone who actually uses a compressor in this fashion deserves to be shot!
You'll notice that some files are much more quiet than others; that's because I didn't adjust the output gain. The ratio was a whopping 50:1, and the threshold a silly -30 dB. Release time was set to 100 ms, for no reason other than "because I said so". Attack times were different. Please note that clicking directly on the links won't work, so you have to right-click and select "save target as", which is fine, because then you can view the waveforms in CE.

This first MP3 is the original file. Actually, it's already been compressed, reverberated, and equalized to a decent state, but it's here just so you get an idea of what I've done to the others.
Right click and select "save target as".

This file has an attack time of 0.50 ms. As you can hear, there is absolutely no attack left in the drums. It's so destroyed, that it sounds like mush. Yes, these settings are silly, but this is what you're doing to your drums when you don't allow sufficient attack time, just to a lesser extent.
Right click and select "save target as".

This file has an attack time of 7.60 ms, and is more to my liking for these particular drums. The transients are left mostly intact, and you still get a sense of punch (although, not with these settings, but you get the point). With controlled settings and this attack time, I was able to compress this file another 6 dB without making it sound too crushed. However, I feel that the original file was good enough...
Right click and select "save target as".

This file has an attack time of 12.90 ms. This attack time is a bit too slow, and you can begin to hear slight pumping artifacts. This is caused by the compression kicking in part-way into the resonance of the drum. Clearly, this is almost as unwanted as the 0.50 ms example.
Right click and select "save target as".

This file has an attack time of 50.00 ms, and things are a bit like the last example, only worse.
Right click and select "save target as".

Under realistic settings, a slower attack time won't even touch the drums. Of course, with the above settings, it just means that the entire thing would likely drop in level, and never return.

I'll gladly answer any questions you have about this.

kylen · Posts: 290

Thanks Post78! I'll look and listen to the examples you sent that illustrate your point. I appreciate your time! Smile

kylen

post78 · Location: USA Posts: 2887

Did it help answer any questions?

kylen · Posts: 290

kylen · Posts: 290

Hi Post78,

OK, first let me put this listening in context to where my head is at today. Following all the good folks on this forum I've been trying to listen to what Bob Katz has to say:

http://www.tcelectronic.com/media/bobkatz.pdf

So bearing in mind that instead of talking in terms of gain reduction he's talking about dynamics preservation and the intelligent use of monitoring and metering if any gain adjustments need be applied.

Since I just downloaded the free VU meters (averaging and PPM) from PSP and don't have time on them yet I'll use the CEP group normalize for peak, average, and RMS metrics and see if this makes sense.

When do dynamics need control? Bob says a program probably needs control when the headroom is greater than 14db or so (average to peak).

On a single instrument like drums taken outside of a song mix that's another thing to. But it is still a program! So let me see what you've sent... Tongue

(Note: I did a group normalize on all the mp3's without limiting and without the equal loudness curve to get the average (rms?), and edit view analyze to get the peak)

"Original.mp3" sounds the best to the ear by itself, naturally as you said you hadn't stepped on it (much). CEP tools would characterize this percussive sound as having a dynamic headroom of about 15.7 db. (-10.4db peak minus average -26.11)

The 0.50 ms mp3 sounds level but very squished but now there is lots of resonance from the kick it sounds like a lo tom! Cool

But the snare is somewhat lost in the hi-hat! The tracks' headroom is at 11.6 db.

The 7.6 ms mp3 sounds like the snare is going to come out of the speaker and poke me in the eye - way too crackin for me. Now the kick is beginning to get lost. This mp3 has a lot of headroom at 23 db but it's too much.

The 12.90 ms mp3 now has the snare breakin up. Its headroom is 22.6 db.

The 50.00 ms mp3 now has the kick coming back in and sounding a lot better but the snare is still crackin too much, in a full mix it would really bust thru though I'll bet. Headroom is about 21 db.

I'm thinking that the 50:1 limiting is a variable to be understood further, the threshold of -30 would also need some assistance by noting the gain reduction lights or metering. Since I have the stats here the gain reduction lights would have said:

original = 0.0 db gain reduction
0.50 ms = 4.1 db gr
7.6 ms = ??
12.90 ms = ??
50.00 ms = ??

But I can't tell what the gain reduction lights would have said where there is an apparant increase in headroom. So I'll just guess that they came on...

Cool experiment Post78, thanks for the resonance thru compression lesson!:D

kylen

post78 · Location: USA Posts: 2887

VoodooRadio · Location: USA Posts: 3971

I created a loop out of the 7.60 file and put it in a mix against guitar and bass. It rocks of course! Thanks for the little clinic Jeremiah. I haven'e tried the UA software, but have used a few of their hardware units. If the software manages to emulate the hardware, then I'm interested. Shy

kylen · Posts: 290

Post78, I forgot to ask. Which compressor are you using. Is Voodoo saying you used a UA plugin like in a UAD-1 DSP ?

That's interesting what Voodoo did, I haven't gotten that far. Put the drumset in a mix and see how it sounds.8)

The difference in peaks that we both have I don't think is a big deal unless I try to borrow your compressor presets and the thresholds wouldn't work anymore. I used the group normalize (without limiting) to get the waves to a similiar loudness level without further stretching them. I could've normalized all of the waves individually to a peak in which case the .50ms squished one would've walked my speaker across the room.

I've been trying to appreciate dynamic headroom lately instead of strictly worrying about 'gain reduction' like the Ozone multiband comp/lim shows me. In other words - do I need to apply dynamics reduction in the first place? Next I want to look at Ozones spectrum some more and see if I can get good average/peak ratios like the PSP compressors give me.

Thanks again! Smile

kylen

VoodooRadio · Location: USA Posts: 3971

jonrose · Location: USA Posts: 2901

post78 · Location: USA Posts: 2887

Kylen:
It was the Waves Renaissance compressor. Since the Renaissance collection emulates older analog gear, its options generally are those found on most any compressor, including Syntrillium's own (I should be careful there, since Synt's comp has so many options that it can sometimes be a bit tricky to use, which can be a great thing). Really though, I wasn't going for anything sounding particularly nice, as is pretty obvious. However, the "original" MP3 used the Ren compressor, Ren EQ, and Ren reverb (in that order, actually), and that's the sound that I actually enjoy with this take (but as mentioned, it was really just to try out the new microphone, and likely won't be kept for anything serious). So that would be a better example of the compressor than the other tracks. You can, of course, use any one of these drum takes in any way you wish, although, I'm not particularly sure why you'd want to...

kylen · Posts: 290

post78 · Location: USA Posts: 2887

kylen · Posts: 290