What a compressor's threshold actually means

Hi. I have a quick question I would appreciate some help with

 

 

If a compressor's threshold is set to -6dB does that mean -

 

 

 

1. Only audio above -6dB will be compressed?

 

 

or

 

 

2. Only sounds above -6dB will trigger the compressor to start working but once the compressor is working ALL audio, regardless of level, will get compressed with the same compression settings (ratio, attack, release etc)

 

 

Which one is it?

 

 

Thanks

Nr 2 is more the correct answer.

When the signal exceeds the threshold it will trigger the compressor's gain controlling circuitry.

Well keep in mind that you're feeding your compressor a single signal, which has only one level – even if that signal is the sum of multiple signals, once it enters the compressor, it's ONE signal. You don't have multiple signals at different levels reaching the compressor.

 

Knowing this...

 

1. Only audio above -6dB will be compressed?

Yes. Once the signal falls below the threshold, it is no longer compressed.

 

2. Only sounds above -6dB will trigger the compressor to start working but once the compressor is working ALL audio, regardless of level, will get compressed with the same compression settings (ratio, attack, release etc)

Yes, because what you call "ALL audio" is actually the (single) audio signal that is fed to the compressor.

Well keep in mind that you're feeding your compressor a single signal, which has only one level – even if that signal is the sum of multiple signals, once it enters the compressor, it's ONE signal. You don't have multiple signals at different levels reaching the compressor.

 

Knowing this...

 

1. Only audio above -6dB will be compressed?

Yes. Once the signal falls below the threshold, it is no longer compressed.

 

2. Only sounds above -6dB will trigger the compressor to start working but once the compressor is working ALL audio, regardless of level, will get compressed with the same compression settings (ratio, attack, release etc)

Yes, because what you call "ALL audio" is actually the (single) audio signal that is fed to the compressor.

 

 

Thanks for that but I think I should have provided more context for this question. Here’s where I’m coming from -

 

 

If you’ve got a mix session with some compression on the stereo bus affecting the entire mix to create some dynamics, the loudest sounds like the snare hits etc will be what triggers the compression. Any quieter sounds that occur immediately after the snare hits will be subject to a swelling/pumping dynamic affect as the compressor releases. I’ve got this happening with one of my mixes at the moment and it sounds good.

 

I’m interested in trying out stem mastering with this track where you split the mix up into 7 or 8 stems (kick, bass, upper drums, synths, vocals etc) and master the stems.

 

 

However, I’m concerned about how the current dynamic pumping/swelling effect of the bus compression (which I like) will be affected if I export stems.

 

 

For example, currently the quieter sounds such as the synths pads, vocals etc are being affected nicely by the swelling compression that is triggered by the loud snare. But if I export the synths as one stem on their own this means the snare will be excluded and it won’t be triggering the compression to act the way it is currently acting in the mix. This will cause the dynamic pump of the synths to be lost, right?

 

 

The same could be said for all the other stems that are affected by the compression triggered by the snare. So I’m afraid that if I export stems, the dynamic interplay between all the instruments will be lost completely.

 

Am I right?

So I’m afraid that if I export stems, the dynamic interplay between all the instruments will be lost completely.

 

Am I right?

Yes.

 

Try this though:

 

- Create a 'ghost' track with your bounced uncompressed mix.

- Set that channel strip to no output.

- Send to a bus, and delete the Aux that Logic automatically creates for that bus.

- Insert a compressor on the main track of each stem.

- Set all the compressors side chain to the ghost track's bus.

So I’m afraid that if I export stems, the dynamic interplay between all the instruments will be lost completely.

 

Am I right?

Yes.

 

Try this though:

 

- Create a 'ghost' track with your bounced uncompressed mix.

- Set that channel strip to no output.

- Send to a bus, and delete the Aux that Logic automatically creates for that bus.

- Insert a compressor on the main track of each stem.

- Set all the compressors side chain to the ghost track's bus.

 

 

Thanks David

 

But would the dynamic interplay between the instruments (and all the other global processing - saturation, imaging etc) in the mix session be maintained exactly if I did the following? -

 

- Turn off all plugins on the stereo bus in the mixing session

- Export the stems from the mixing session

- Import the stems into the mastering session

- Copy the plugin chain that was on the stereo bus in the mixing session and paste it on to the stereo bus in the mastering session?

 

 

Thanks

Assuming your stems levels are reasonable, and you follow the same signal routing, it should give you the same result, yes.

However, "all that being said," will the recipient of "that stem (sic ...) collection" love you, or hate you?

 

To my way of thinking, when you send a project downstream as "stems," your goal is to provide the downstream engineer with: "the inputs to your mix."

 

If you want the engineer to achieve a particular effect that has become near-and-dear to your ears, then I suggest that you send "your mix" as supplemental information, along with technical details of how you set it up, and exactly ("pumping ...") what pleases you about it.

 

To my way of thinking, if you instead modulate your stems (with anything at all ...), what you have actually done is to inject a noise element – your "pumping" – into the signal. Because you have now done so at the input side of the engineer's signal-chain, whereas it originally occurred at the output side of yours, you have just gifted your downstream colleague with: "a noisy signal." (And noise, once injected, cannot be removed. It can only accumulate, or multiply.)

 

Whereas, if you provided "exactly the same inputs that you had," along with an example of "what tickled your ears" and technical details for obtaining it, the engineer could probably make "the customer" (you ...) more-than happy. You have now presented "the signal" and "the (beautiful ...) noise" separately.

Thanks Mike. I will be mastering it myself though in this case.

... then David's suggestion would of course be an expedient way. (Suh-prize!) Let the stems be pristine, and use a previously-prepared reference track that no one actually hears (the "ghost") to provide side-chain inputs during your mixing process. Just take care that the pulsing modification is applied to any particular signal only once: don't pulse what has already once been pulsed.

1. Only audio above -6dB will be compressed?

Yes. Once the signal falls below the threshold, it is no longer compressed.

Actually the signal will still be compressed even though it falls below e.g. -6 dBFS, unless the compressor has an infinitely quick release time.

 

What falls below -6 dBFS is the input signal that the detection circuit in the internal side-chain is listening to in order to determine the gain reduction.

 

The output signal, which is what you're hearing, will still be compressed until the actual gain reduction is at 0 dB, which at this stage is mainly due to the release time.

 

I don't like to use the word "stage" because there really is no attack and release stage in compression per se, even though that's the common explanation, and one that's even espoused on Wikipedia. In reality regular dynamic compression is a contant flux between turning up and turning down even when it appears to be at a steady hold, which is why you get intermodulation distortion.