Why do compressors react differently when tracked through?

[demel]

The drum room mics have a very different sound when I record through them into ProTools with lots of squish rather than compressing them after the fact.

Why is that?

Its not just a software vrs hardware thing either as running the signal back though a hardware unit doesn't sound the same either.

[MarkRB]

Yeah that's one of those mysteries I've always wanted to know the answer to.

I always prefer the sound of a mic through a compressor than a tape return through the same box, but there is no real good reason for it. I've always imagined it was a transient thing, or it could all just be psychoacoustics?

[lespaul666]

It's the A/D process,(sorry MarkRB, missed your comment of "off tape") and the character of the analog recording process itself. It's going to sound different once it's recorded, regardless if it's analog or digital. Even high sample rates at 24-bit. The sound is never going to be completely true to the direct input signal going through a compressor.

[MarkRB]

Originally Posted by lespaul666

It's the A/D process.

That I would believe if I hadn't noticed the same thing off tape.

[nickynicknick]

because once it hits your DAW or tape it sounds different...once a sound is converted....to either magnetic pieces on tape or ones and zeros n your DAW it's very different than when it came into the cable from the source....BTW I think it's a beautiful thing that this exists because it forces you to make immediate decisions while tracking.

Nick

[Zep Dude]

Originally Posted by MarkRB

That I would believe if I hadn't noticed the same thing off tape.

Actually, I tend to prefer compression after tape. I like the tape to take the first "hit" and do it's thing. I also found with a stereo mix that uses a bus compressor, that when going to 2 track I actually like printing the mix without the bus comp and then adding the bus comp back in on the way to the AD converter (post tape).

[Zep Dude]

Oh, in answer to your question, as someone else said, the storage medium changes the sound. It's never as dynamic once it goes to digital or to tape.

[synthoid]

It wouldn't surprise me if it were next to impossible to get D/A converters to recreate exactly the signal strength and dynamics of the original analog signal that hits the compressor when you track through it.

-synthoid

[Sigma]

Originally Posted by synthoid

It wouldn't surprise me if it were next to impossible to get D/A converters to recreate exactly the signal strength and dynamics of the original analog signal that hits the compressor when you track through it.

-synthoid

with 16,777,216 levels of resolution in a 24 bit converter i doubt that is true

[bdubya]

If the player is hearing the compression, they're gonna play a little different than when they're hearing themselves non-compressed,but then you compress that performance with the same settings. Its sort of inspiring to hear that familiar compressed drum sound in the cans when you're playing, IMO. Not the fanciest techno answer, but probably valid . . .

[nandoanalog]

Originally Posted by bdubya

If the player is hearing the compression, they're gonna play a little different than when they're hearing themselves non-compressed,but then you compress that performance with the same settings. Its sort of inspiring to hear that familiar compressed drum sound in the cans when you're playing, IMO. Not the fanciest techno answer, but probably valid . . .

Very valid. If a drumer is listening to a compressed mix, he will play the drums in a way to make it sound as good as possible with those compressor settings.

[nickynicknick]

Originally Posted by Sigma

with 16,777,216 levels of resolution in a 24 bit converter i doubt that is true

different thread dude....oh and..... you're wrong.

[Sigma]

Originally Posted by nickynicknick

different thread dude....oh and..... you're wrong.

2 to the 24th power = 16 777 216...and that is the number of level increments a 24 bit processor can express

[Francis Vaughan]

Theoretically the dynamic range of a 24 bit channel is 147.5dB. But there isn't a converter on the planet that can actually achieve that. (It spans the range from lowest discernible sound pressure up to permanent hearing damage.) Most converters would be lucky to be good to 20 bits. Which is still a pretty good achievement.

The effects of the anti-alias and reconstruction filters in the AD-DA chain probably need to be factored in to this discussion as well. Not to mention any FIR filters in the digital domain.

Trivial things, such as a compressor responding to subsonic or ultrasonic components that are missing from the recorded signal could be an issue as well.

[SighBorg]

The effects of the anti-alias and reconstruction filters in the AD-DA chain probably need to be factored in to this discussion as well. Not to mention any FIR filters in the digital domain.

Trivial things, such as a compressor responding to subsonic or ultrasonic components that are missing from the recorded signal could be an issue as well.

Pretty much exactly what I was going to post as well.

Also, by filtering off the ultra-sonics you (as far as I'm aware) are removing heterodyne interactions which can extend down into the ranges more relevent to a compressor's rms/peak trigger.

I never thought it was some mystical voodoo... Plug a synth directly to a speaker and listen and then monitor it through a DAW and listen. It doesnt sound the same to the ears so why should a compressor react the same? They're both listening to the same thing.

[Long_Shaded_Eye]

Probably because the compressor is doing something to a part of your that isn't here anymore after the AD. (i'm working in digital anyway).

I'm sure it has something to do with the harmonics.
Peace.
/Nick.

[robd]

I'm not trying to write off the potential for minor effects from a lot of these sources people are suggesting, but I think bdubya hit the nail on the head with the overriding factor. I certainly notice that I play the bass differently when playing through a comp, and there's no way to recreate that after the fact.

[Sigma]

Originally Posted by Francis Vaughan

Theoretically the dynamic range of a 24 bit channel is 147.5dB. But there isn't a converter on the planet that can actually achieve that. (It spans the range from lowest discernible sound pressure up to permanent hearing damage.) Most converters would be lucky to be good to 20 bits. Which is still a pretty good achievement.

The effects of the anti-alias and reconstruction filters in the AD-DA chain probably need to be factored in to this discussion as well. Not to mention any FIR filters in the digital domain.

Trivial things, such as a compressor responding to subsonic or ultrasonic components that are missing from the recorded signal could be an issue as well.

even it their where 1/10 those levels available the gain changes would sound the same to our ears

[synthoid]

Originally Posted by Sigma

even it their where 1/10 those levels available the gain changes would sound the same to our ears

Are you trying to prove by arithmetic that all D/A converters are perfect, or that they are all equivalent to one another? You're ignoring that there's an analog circuit involved; the converter isn't merely doing arithmetic on the incoming bits.

---

It's interesting how some singers will hear a compressor and start using it as part of their mic technique, e.g., sing more freely in loud passages.

-synthoid

[Sigma]

Originally Posted by synthoid

Are you trying to prove by arithmetic that all D/A converters are perfect, or that they are all equivalent to one another? You're ignoring that there's an analog circuit involved; the converter isn't merely doing arithmetic on the incoming bits.

---

It's interesting how some singers will hear a compressor and start using it as part of their mic technique, e.g., sing more freely in loud passages.

-synthoid

the post i responed to was about relative values of LEVEL in vs relative values of LEVEL out and how they correlate..i know didlly about converter arithmatic ..but can graspe fact that at 24 bit the resolution of volume changes is pretty damn past any incremental stuff the human ear can hear

[jimmyjimjim]

A few factors come to mind:

-Impedance less than optimal between your compressor and AD/DA, depending on ITB, or OTB via insert this can be a factor, and known to be one specially on digidesign gear.
Also most inserts are ubalanced, which you have to make up for in your gainstaging.

-Musicians play differently if they hear the compression while playing.

-Tape: Tape pushed hard is a multi-band compressor in itself, the response of the "real" compressor will vary according to being before and after that.

-Latency issues. If send and return from in the box. The Latency will NEVER be 100% perfect, even 1 sample accuracy is not always enough when you are dealing with a signal that has been reconstructed into analog and then converted back to digital.
Sampling at 44,1, you can get a 180 degree phase shift at 20k even with one sample accuracy.

One thing that does not, however even enter this discussion, is this bullshit about there being so-and-so many volume steps(but not enough blah blah.) and filter this and that in 24bits. It's magnitudes more subtle than all the factors above, and I'm sure we can come up with more.

The single most important factor though, imo, is the differences in how we listen, and how we focus in a tracking situation vs. a mixing situation.

[ryanojohn]

Your compressor is often affecting signal over 20kHz, and signal below 20Hz, depending on the quality and bandwidth of your pres and signal path. This undoubtedly introduces harmonic artifacts in the audible spectrum...