Software EQ matching

[Xenu]

Recently, I was making a pass at creating a "custom" version of Frank Zappa's "You Are What You Is." The current stock CD sounds great, but contains a slight edit for content which I wanted to see if I could better (or at least equal). By working with the old EMI CD, I was in fact able to make something I thought was perfectly acceptable.

Problematically, the EMI CD is EQed very differently from the current disc; it's tinny, for one, and is not nearly as rich-sounding as the current master. Play as I might with EQ settings, I never got something which sounded right, although I felt that I should have been able to get a lot closer than I did.

This problem also occured a few months ago while I was trying to edit together a composite version of a track by the Majestic Four. Two *slightly* different masterings of the same tune, and yet I couldn't get the EQs of the different segments to gell.

Is there something I'm missing? Some way to get the EQ close enough without the guesswork? Or is this simply why I'm not an audio engineer?

[Dob]

Unfortunately, most EQ is guesswork, albeit somewhat "educated."

In my experience, most of the inscrutable EQ problems crop up when I'm trying to fill a hole by subtracting EQ, or vice versa. If there's a "hole" in the sound, working your way through the entire spectrum, trying a cut at every frequency, won't do the trick.

To further compound the problem, your EMI CD may have a "hole" in the low bass and high treble along with a "peak" at, say, 3K. Trying to correct really screwed up tracks with EQ is like trying to pick a lock. Some open up right away, others are stubborn.

For your "tinny" CD, I would start with concentrating on the vocals and limiting your adjustments in the 2-6K range (probably a cut). Then work on the treble, above 6K...you'll probably need a boost here. Finally, fix the deep bass (below 50hz)...you'll need plenty of boost in this area. If you need added richness, boost from 120hz to about 400hz evenly -- about 1/2 db should suffice. Anything more will probably sound muddy.

I prefer very "notchy" boosts and cuts (unless I'm correcting a gradual bass or treble rolloff) and I don't like the "slider" EQ programs that mimic a real EQ. The "parametric" EQs, where you build your own curve using data points, sound and work much better IMO.

[Xenu]

Problematically, it's very hard to do an adjustment and then immediately hear the "before" and "after" together. With the Zappa discs, the EQ is off to a severe enough extent that I really wouldn't even know where to begin.

I've put a sample up at http://www.lukpac.org/~handmade/daotest.mp3 . The Ryko comes first, and then it switches at each repetition (barring the one right before the fade).

-D

[lukpac]

Doesn't your EQ have a "compare" button, so you can switch it in and out?

[Dob]

Doesn't your EQ have a "compare" button, so you can switch it in and out?

You can also set up a second "flat" EQ curve and use that as your "before", switching back and forth.

Or, you can play the EQed track in sync with the original and do an A/B comparison.

IMO it is essential to have your computer hooked up to your hifi so you can do proper comparisions. I have a "hifi link" that uses a USB port and connects directly to my hifi through RCA jacks. It also bypasses my sound card. On top of that, it's cheap (around $40).

David, when I have more time I'll have a listen to your sample and post my opinion.

[krabapple]

Doesn't your EQ have a "compare" button, so you can switch it in and out?

You can also set up a second "flat" EQ curve and use that as your "before", switching back and forth.

Or, you can play the EQed track in sync with the original and do an A/B comparison.

IMO it is essential to have your computer hooked up to your hifi so you can do proper comparisions. I have a "hifi link" that uses a USB port and connects directly to my hifi through RCA jacks. It also bypasses my sound card. On top of that, it's cheap (around $40).

? If it bypasses the card..waht is it actually sending to the receiver via RCA jacks?

David, when I have more time I'll have a listen to your sample and post my opinion.

What about doing a frequency profile of the same segment each each, using Cool Edit or somesuch, then trying to match them based on that? Has anyone ever tried it?

[Dob]

If it bypasses the card..waht is it actually sending to the receiver via RCA jacks?

I honestly don't know how the damn thing works, but you can check it outhere.

[Rspaight]

Looks like it basically *is* a USB sound card (which Windows supports natively), without any sort of amplifier or inputs. (They sell the USB-based "INport" for analog inputs.) Interesting. In theory, I guess, you wouldn't *need* a sound card with one of these.

Ryan

[Dob]

I can vouch for the sonic transparency of the hifi link...it's not perfect but it's much better than my (crappy Soundblaster) card. I have a hard time telling the difference between the wav file played through the link and the original CD in an A/B comparison.

[Dob]

What about doing a frequency profile of the same segment each each, using Cool Edit or somesuch, then trying to match them based on that? Has anyone ever tried it?

What may work even better is an inverse compare, and then a frequency analysis of the resulting differences. It would be a good starting point, anyway.

[Xenu]

What about doing a frequency profile of the same segment each each, using Cool Edit or somesuch, then trying to match them based on that? Has anyone ever tried it?

That's something I was thinking of doing. It's harder than it sounds, though. You know what you want your "after" picture to look like, but getting there is still something you have to trial-and-error out.

What may work even better is an inverse compare, and then a frequency analysis of the resulting differences. It would be a good starting point, anyway.

That only works if the material in question derives from the same digital source. In this case, we're dealing with two different digital transfers, so there's no sync.

[Dob]

That only works if the material in question derives from the same digital source. In this case, we're dealing with two different digital transfers, so there's no sync.

That did cross my mind. But perhaps if you compared a short snippet -- one second, for example -- the difference in sync wouldn't skew the results so much as to make them unusable.

[lukpac]

It doesn't matter how long the snippet is - if they aren't from the same source, they won't be in perfect sync, period.

[Dob]

It doesn't matter how long the snippet is - if they aren't from the same source, they won't be in perfect sync, period.

I understand that. But why do they have to be in perfect sync?

[lukpac]

If they aren't in perfect sync, you're not just comparing EQ anymore.