Using an expander to counter a compressor --

Topic: Using an expander to counter a compressor -- (Read 7292 times) previous topic - next topic

0 Members and 1 Guest are viewing this topic.

Using an expander to counter a compressor --

2006-01-13 00:06:44

Let's say I have RHCP's Californication and I want to "remaster" it, without having the original tapes. Will using a dynamics expander "correct" what bad mastering was done, to an extent, or is the album good for the trash in any case?

I was just thinking that with so many overloud albums, this might be a recommendable technique to counter some of the mastering engineer's work, no?

Thanks for your replies,
Shade

Using an expander to counter a compressor --

Reply #1 – 2006-01-13 00:15:18

You might make something that sounds better to you but, somewhat like lossy compression (as an analogy), you can not get back to what you (or the recording engineer, actually) started with.

Using an expander to counter a compressor --

Reply #2 – 2006-01-13 01:01:58

If all that was done to the master was compression, then you MIGHT have been able to undo some of the damage IF you really knew how to use an expander.
However, what has occurred at the mastering stage of almost every album released in the last 10 years or so is not just compression but limiting, and that's a whole new beast entirely.
And no, I'm sorry to say that no amount of tweaking with an expander is going to undo that damage.
Hopefully, 20 years from now, someone will wake up to all the damage that's been done and release "audiophile" versions of all those albums without the hyperlimiting that's currently being used.... and this will eventually happen. Know why? 'Cause the record companies will be suffering from lost revenue due to file sharing and so on, and they'll be looking for every conceivable avenue to squeeze more money out of the music-consuming public.

Using an expander to counter a compressor --

Reply #3 – 2006-01-13 01:24:41

Quote

Audio level compression
From Wikipedia, the free encyclopedia.

Audio level compression, also called compression or limiting...

Nothing can be done? Are you sure?

Using an expander to counter a compressor --

Reply #4 – 2006-01-13 01:41:50

Quote

Quote
Audio level compression
From Wikipedia, the free encyclopedia.

Audio level compression, also called compression or limiting...

Nothing can be done? Are you sure?
[a href="index.php?act=findpost&pid=356685"][{POST_SNAPBACK}][/a]

The expander will only enhance the peaks at a specified frequency, but it won't do any kind of miracles on a full track. You'd have to run it on each of the tracks which have been mixed down by the audio engineer and it wouldn't yet compensate for the agressive limiting. You may also try a transient processor like Digital Fish Phones "Dominion(VST)" or Voxengo's "Transmoder"(VST as well). In any case what's been lost in the limiting process is lost forever.

This is the most annoying aspect of the so called "loudness race". The best you can do is to petition the group or the label and ask for a remaster.

Using an expander to counter a compressor --

Reply #5 – 2006-01-13 01:48:09

Quote

Quote
Audio level compression
From Wikipedia, the free encyclopedia.

Audio level compression, also called compression or limiting...

Nothing can be done? Are you sure?
[a href="index.php?act=findpost&pid=356685"][{POST_SNAPBACK}][/a]

Positive.
Sorry.

Using an expander to counter a compressor --

Reply #6 – 2006-01-13 01:54:21

Quote

The expander will only enhance the peaks at a specified frequency,

That's slightly misleading as expanders don't work on peaks, they work on low level signals, but by expanding the lower levels, it gives the IMPRESSION of improving the peaks.
And as for "at a specified frequency", that might be true IF it's a multiband expander. Most aren't. Most work on full spectrum audio.
And to answer the previos poster about compression and limiting being the same thing... no, they aren't.
Compression pulls loud peaks down by a certain amount (as set by the "ratio" on the compressor), whereas a limiter simply chops the tops off clean at the threshold. A limiter leaves the audio with square tops. Open a top 40 single in an audio editor like Audition and zoom in on one of the loud peaks. You'll see what I mean. They're square. And that kind of damage cannot be undone.

Using an expander to counter a compressor --

Reply #7 – 2006-01-13 02:17:23

If you use an expander on the already-processed, signal, you will actually end up with a signal with less dynamic range, not more. Most expanders work by "turning up" the volume unless a given threshold is reached. So, less dynamics, not more.

[span style='font-size:8pt;line-height:100%']Edited 1-14-05 @ 11:31am[/span]

Using an expander to counter a compressor --

Reply #8 – 2006-01-13 02:52:41

Quote

If you use an expander on the already-processed, signal, you will actually end up with a signal with less dynamic range...

Sorry, I disagree with that. The job of an expander is to EXPAND dynamic range. You set a threshold and any audio which falls BELOW that threshold is turned DOWN, thus increasing the difference in volume between the loudest bits and the quietest bits. It's like a subtle version of a gate. A gate will cut off any signal that falls below the threshold, while an expander will merely turn it down a bit further (according to the set ratio).

Quote

Most compressors work by "turning up" the volume unless a given threshold is reached[a href="index.php?act=findpost&pid=356695"][{POST_SNAPBACK}][/a]

Sorry. Wrong again.
A compressor works on audio which EXCEEDS a set threshold. If a peak exceeds that threshold, the compressor will reduce the excursion of that peak according to the ratio set. The "turning up" that you refer to is actually "makeup gain" which is applied to the WHOLE signal AFTER compression has occured.

Using an expander to counter a compressor --

Reply #9 – 2006-01-13 11:06:51

Firstly, I agree with Audio2U in large part, but in theory there is something that could be done. (I mention it purely out of academic interest - it serves no practical purpose):

There are an infinite number of possible compression scenarios which could have resulted in the audio to be "restored". Having the end result doesn't tell you anything about what compression was actually applied. That said, expansion CAN undo the compression, but only if you know all of the following:

1. The linear gain which was applied after compression (the "makeup gain" referred to by Audio2U).
2. The threshold level above which compression was applied.
3. The degree of compression applied.
4. The attack and release times of the compressor.
5. Probably some other parameters that I've forgotten about.

If you know all of these things, then in theory it would be possible to set up an expander to undo the effects of the compression, although you would suffer from a slight increase in noise floor because the compressor, in squashing the dynamic range, effectively decreases the bit depth of the signal.

Of course nobody is ever going to supply the necessary data. Quite probably it was never even noted down at the time of mastering. In some cases I doubt if the mastering engineer even knew what they were - he would have just twiddled the knobs until it was as loud as required.

One further problem: if any limiting was applied, then you're stuffed: that simply destroys huge amounts of information. So that rules out any possibility of restoring the dog's dinner clip-fest that is Californication (and any number of other recent CD releases).

Using an expander to counter a compressor --

Reply #10 – 2006-01-13 12:46:40

Bingo.

Using an expander to counter a compressor --

Reply #11 – 2006-01-14 02:04:51

I tried that. Several expansion settings using cool edit's "dynamics processing". Looks like it doesn't work because I don't know the settings used on mastering, so I confirm what cliveb said above.
In some few cases, where only compression and no limiting has been used, it's possible to make the differences between the peaks bigger because there's still something left of the original hugh peak differences. It may sound "better" but probably more different to the original than the master that you started with.

Quote

A limiter leaves the audio with square tops. Open a top 40 single in an audio editor like Audition and zoom in on one of the loud peaks. You'll see what I mean. They're square. And that kind of damage cannot be undone.
[a href="index.php?act=findpost&pid=356690"][{POST_SNAPBACK}][/a]

In my opinion, the difference between limiter and compressor is that a limiter uses infinite compression ratios, it sets the peaks to the exact same level, while a compressor does not. Example: Peaks at 0dB and -10dB are set to -9 and -12 by a compressor but a limiter sets it both to -11.
Those "square tops" that you are talking about, audio2u, are clipped peaks. Try cool edit's (or Audition's) hard limiter. Limiters don't create flat lines (zoom in).

Using an expander to counter a compressor --

Reply #12 – 2006-01-14 02:31:50

Actually wouldn't compression actually increase the effective bit depth? I mean, if you were able to reverse it, the reversal would often attenuate enough to give a few extra bits.

I agree that in the general case this is a basically impossible problem. However, if a specific case were provided for encoding/decoding, then that could be promoted to mastering studios. "You can still use a decent compressor and users can totally undo it if they want to"

Of course, this is all just reinventing HDCD, sooooooo......

Using an expander to counter a compressor --

Reply #13 – 2006-01-14 13:18:43

Quote

In my opinion, the difference between limiter and compressor is that a limiter uses infinite compression ratios, it sets the peaks to the exact same level, while a compressor does not.

Yes, that is correct.

Quote

Example: Peaks at 0dB and -10dB are set to -9 and -12 by a compressor but a limiter sets it both to -11.

I'll go along with that in theory. It all depends on the settings of both the compressor and limiter.

Quote

Those "square tops" that you are talking about, audio2u, are clipped peaks.

Yes, I realise that.

Quote

Limiters don't create flat lines (zoom in).
[a href="index.php?act=findpost&pid=356906"][{POST_SNAPBACK}][/a]

They do if you hit 'em hard enough!
Remember, the shape of the output peaks will be a function of both the threshold AND the release time of the limiter. So, set the threshold low enough and the release time long enough, and my word you WILL get flat tops!
And what the original poster was talking about was RHCP's "Californication" which is a well known example of an album squashed in mastering.

Using an expander to counter a compressor --

Reply #14 – 2006-01-14 13:29:19

Quote

Actually wouldn't compression actually increase the effective bit depth?

No. Let's assume that the source tape is 16 bit (it wouldn't be, it would actually be 24 bit, but let's pretend, ok?) which gives it 96dB of dynamic range.
You then compress the top 26dB of that.
You've just wasted the top 4.5 bits of DR (each bit gives you 6dB of DR).
You apply makeup gain of 26dB to bring the peaks up to 0dBFS, and guess what?
You had nothing at the bottom end of the scale so we introduced 4 bits of noise to the floor!
Woohoo.
Not.

Quote

I mean, if you were able to reverse it, the reversal would often attenuate enough to give a few extra bits.

Interesting theory.
I won't say definitively that it wouldn't work, but your chances of finding the right numbers are pretty small.

Quote

I agree that in the general case this is a basically impossible problem. However, if a specific case were provided for encoding/decoding, then that could be promoted to mastering studios. "You can still use a decent compressor and users can totally undo it if they want to"

Of course, this is all just reinventing HDCD, sooooooo......
[a href="index.php?act=findpost&pid=356909"][{POST_SNAPBACK}][/a]

You honestly think mastering engineers WANT end users to be able to undo their work?
Do you think the record companies want you to be able to undo it either?
Not bloody likely!

Using an expander to counter a compressor --

Reply #15 – 2006-01-14 13:33:30

Quote

Actually wouldn't compression actually increase the effective bit depth? I mean, if you were able to reverse it, the reversal would often attenuate enough to give a few extra bits.[a href="index.php?act=findpost&pid=356909"][{POST_SNAPBACK}][/a]

If you compressed an analogue signal *before* digitising, then expanded afterwards, I suppose that would effectively increase the bit depth (provided when you exapnd you increase the word length). If you think about it, this is how dbx noise reduction works (albeit in the analogue domain).

But if you've got a digital recording and then compress it, you reduce the difference between loud and quiet samples, which then uses fewer bits to encode. You are effectively throwing away the bottom N bits of resolution. When you expand again, the difference in signal level between loud and quiet samples is less well defined, resulting in higher quantisation noise.

Here's and example to illustrate. Suppose you've got two samples with the numeric values N and N+3 and that compression changes them to N and N+1. Now take another pair that is initially N and N+2. The same compression settings would require them to be N and N+0.67. But fractional sample values can't be stored, so it gets rounded to N and N+1. Now we see that both pairs are the same: the subtle difference in relative levels between the samples has been lost, and on expansion it cannot be recreated: both pairs of samples would end up at N and N+3. This inaccuracy represents an increase in quantisation noise.

Using an expander to counter a compressor --

Reply #16 – 2006-01-14 17:38:37

Quote

Sorry, I disagree with that. The job of an expander is to EXPAND dynamic range. You set a threshold and any audio which falls BELOW that threshold is turned DOWN, thus increasing the difference in volume between the loudest bits and the quietest bits. It's like a subtle version of a gate. A gate will cut off any signal that falls below the threshold, while an expander will merely turn it down a bit further (according to the set ratio).

Let's be specific: what expanders in particular are we talking about? Are we discussing plugins for Pro Tools, Adobe Audition, pro audio analogue outboard gear, what?

I'm a sound engineer, and every expander I've used that lives in a rack works by boosting the signal unless the signal reaches that threshold on its own. They're usually used in conjunction with a compressor to have a very limited dynamic range on something. They're often used in broadcast situations where someone doesn't want to have worry about riding faders.

Quote

Most compressors work by "turning up" the volume unless a given threshold is reached

I mis-typed up there. I've corrected my post.

Using an expander to counter a compressor --

Reply #17 – 2006-01-14 22:52:53

Quote

Are we discussing plugins for Pro Tools, Adobe Audition

as for me, yes

Quote

I'm a sound engineer, and every expander I've used that lives in a rack works by boosting the signal unless the signal reaches that threshold on its own. They're usually used in conjunction with a compressor to have a very limited dynamic range on something. They're often used in broadcast situations where someone doesn't want to have worry about riding faders.

I believe that, that's exactly what I would epect in the radio studio, based on what I hear.

Quote

Quote
Most compressors work by "turning up" the volume unless a given threshold is reached

I mis-typed up there. I've corrected my post.

Sorry, I'd say with expander instead of compressor it's even more wrong. The expander increases the loudness of quiet parts and affects only them, thus decreases the overall dynamic range, thus it's a compressor.

Ever played around a bit with those Pro Tools ? In cool edit/Audition the dynamics processor has tons of options, I can create an input-output curve that leaves every audio below (say) -10dB unaltered but amplifies every peak that exceeds that limit. That's what I call a (real) expander. (Be aware of possible clipping in 16bit.)

ps: I like your avatar, being aware of the volume knob is often a good idea!

edit: corrected grammer

Using an expander to counter a compressor --

Reply #18 – 2006-01-14 23:22:14

Ah-hah! Having just read Precisionist's last post, I can now see where the confusion is coming from.
Flattop100, ok, I'll consent that you were right as long as you'll agree that I was right, too!
I was referring to a downward expander (which I thought got mentioned earlier in this thread, but maybe I'm imagining things).
The kind of expander you're referring to I've only ever encountered in plugins. I too am a working engineer (I work in studios, not sure if you do the same or live stuff?), and all the hardware expanders I've seen have been of the "downward expansion" variety.
So, perhaps for the less-technical readers we need to explain (if it hasn't been covered thoroughly enough yet) the difference between the kind of expander Flattop100 was referring to, and the "downward expander" that I was referring to.
Precisionist, your comment The expander increases the loudness of quiet parts and affects only them, thus decreasing the overall dynamic range, thus it's a compressor. is not entirely correct. It is correct in the sense that it will reduce the dynamic range, thus giving the appearance of working like a compressor, but the method of achieving that reduction in dynamic range is totally different to the way a compressor would achieve it. A compressor works from the top down, where the expander works from the bottom up.
Apologies if I'm confusing anyone!

Using an expander to counter a compressor --

Reply #19 – 2006-01-15 00:35:30

Quote

But if you've got a digital recording and then compress it, you reduce the difference between loud and quiet samples, which then uses fewer bits to encode. You are effectively throwing away the bottom N bits of resolution. When you expand again, the difference in signal level between loud and quiet samples is less well defined, resulting in higher quantisation noise.

Here's and example to illustrate. Suppose you've got two samples with the numeric values N and N+3 and that compression changes them to N and N+1. Now take another pair that is initially N and N+2. The same compression settings would require them to be N and N+0.67. But fractional sample values can't be stored, so it gets rounded to N and N+1. Now we see that both pairs are the same: the subtle difference in relative levels between the samples has been lost, and on expansion it cannot be recreated: both pairs of samples would end up at N and N+3. This inaccuracy represents an increase in quantisation noise.
[a href="index.php?act=findpost&pid=357012"][{POST_SNAPBACK}][/a]

Ah, yeah, I should have realized that. An shorter way to say it is that "mu-law is not a free lunch". That is, an expansion algorithm is very similar to a logarithmic encoding in that the upper bits pack more data.

Notice