Gang all the Gain controls together

[JackA]

When mixing many recorded tracks and I'm satisfied with the mix, but the amplitude peaks (VU) are too high, it would be nice to be able to gang all the Gain controls together - moving one moves them all. I was doing this with the Amplify feature (selecting no particular track), but too much modification can be harmful. Great program! Nice work! Thank You!

[kozikowski]

This may be buried in another feature request. The two tools, Amplify and Normalize work wherever they're pointed, but I don't think either of them will prevent a composite show from exceeding 0dB on export. I guarantee that's come up multiple times. I believe there's a painful work-around, but no simple tool for a problem that everybody with a multi-track show is going to run into.

Something like Mix and Render to a separate track and then process that separate track while the show is still in 32-floating format. I only got pieces of this.

Koz

[audiolover]

When mixing many recorded tracks and I'm satisfied with the mix, but the amplitude peaks (VU) are too high, it would be nice to be able to gang all the Gain controls together - moving one moves them all. I was doing this with the Amplify feature (selecting no particular track), but too much modification can be harmful. Great program! Nice work! Thank You!

Brilliant idea, i was thinking the same although

[steve]

How about if the Mixerboard had a "master" slider that ganged the sliders of all selected tracks so that moving the master slider by, say -4 dB, would cause the Gain sliders of all selected tracks to move by the same amount (-4 dB)?

but too much modification can be harmful

If the tracks are 32 bit float format, then it's almost impossible to cause damage through repeated amplification - you would need to apply really excessive amounts of amplification (like +/- 10000 dB) to cause damage.

[Robert J. H.]

How about if the Mixerboard had a "master" slider that ganged the sliders of all selected tracks so that moving the master slider by, say -4 dB, would cause the Gain sliders of all selected tracks to move by the same amount (-4 dB)?

but too much modification can be harmful

If the tracks are 32 bit float format, then it's almost impossible to cause damage through repeated amplification - you would need to apply really excessive amounts of amplification (like +/- 10000 dB) to cause damage.

That's a little bit high, Steve...

The highest (integer) number that can be displayed with 32-bit float (and is still 1 higher as the previous one) is 2^53.
In dB, this is 319.092 amplification.
Linearly, the number is:
9007199254740992.0

For the nyquist prompt (debug):

Code: Select all

(setf bignum (expt 2.0 53.0))
(setf *float-format* "%f")
(print (1- bignum))
(print bignum)
(print (1+ bignum))
(print (linear-to-db bignum))

numbers that exceed this limit are quantized into a certain range.
Note that this begins even earlier for the fractional parti.e. the limit for 0.5 steps is accordingly lower.

The same happens at the other end of the scale (1/num).
Amplifying by -6 dB and 6 dB can already delete 1 bit (It's actually 6.0206 dB).
However, this is a really tiny amount and far below any noise floor of your gear. Apart from that, this bit will be "refilled" by another action such as filtering or compression.

In general, amplification and normalisation won't be harmful to your audio.
A highpassfilter will in contrast cause much more damage.
The reason is that it is implemented as infinite impulse response. Thus, any quantization errors add up over the whole length of the audio--still, not worth worrying about.

[steve]

That's a little bit high, Steve...

I was referring to an amount of amplification that will cause damage.

A very low level signal (say around -500 dB), may be amplified to well over +500 dB (1000 dB of amplification) almost perfectly.
10000 dB was just an example of an extreme amount of amplification that is beyond the capabilities of 32 bit float format - the point being that for "normal" or even "very large" amounts of amplification, providing that the audio format is 32 bit float, amplification is totally harmless. You have to go into the realms of stupid amounts of amplification to cause noticeable damage. Being that the limit of the Amplification effect is +/- 50 dB, it is not easy to produce the amounts of amplification necessary to cause damage.

[Gale Andrews]

I'll make a note of the two "votes" for this.

Gale

[steve]

I'll make a note of the two "votes" for this.

Would you make that 3 votes. (one from me).
I'd suggest this as a way of implementing it: http://forum.audacityteam.org/viewtopic ... 73#p271073

[Gale Andrews]

I'll make a note of the two "votes" for this.

Would you make that 3 votes. (one from me).
I'd suggest this as a way of implementing it: http://forum.audacityteam.org/viewtopic ... 73#p271073

I believe you already voted, but not explicitly for the solution of adding a master slider to Mixer Board. Still, that can be added as a subvote without changing the total votes.


Gale

[Robert J. H.]

That's a little bit high, Steve...

I was referring to an amount of amplification that will cause damage.

A very low level signal (say around -500 dB), may be amplified to well over +500 dB (1000 dB of amplification) almost perfectly.
10000 dB was just an example of an extreme amount of amplification that is beyond the capabilities of 32 bit float format - the point being that for "normal" or even "very large" amounts of amplification, providing that the audio format is 32 bit float, amplification is totally harmless. You have to go into the realms of stupid amounts of amplification to cause noticeable damage. Being that the limit of the Amplification effect is +/- 50 dB, it is not easy to produce the amounts of amplification necessary to cause damage.

Quite so.
I just wanted to clarify that 10000 dB is a very huge number:
10000 dB = 10^500
The atoms in the universe do only add up to 10^80, thus each particle could hold another universe whose particles hold another universe... 5 times over.

It's remarkable how concise such multiplications/divisions can be expressed with the help of the dB scale.
Even shorter would a scale of factorials (1*2*3...*n) be:

Code: Select all

Amplification	Factorial	Linear Factor
0 dB	1!	1
6 dB	2!	2
16 dB	3!	6
28 dB	4!	24
42 dB	5!	120
57 dB	6!	720
74 dB	7!	5040
92 dB	8!	40320
111 dB	9!	362880
131 dB	10!	3628800
152 dB	11!	39916800
174 dB	12!	479001600
196 dB	13!	6227020800
219 dB	14!	87178291200
242 dB	15!	1307674368000
266 dB	16!	20922789888000
291 dB	17!	355687428096000
316 dB	18!	6402373705728000
342 dB	19!	121645100408832000
368 dB	20!	2432902008176640000

Apart from those numerical thought jumps, I certainly support the idea of a master gain slider.
We've already discussed this in another feature request.
The volume handling is certainly tricky. There are already a lot of "volume knobs" in the recording/playback/export chain.
In principle, I'd like to see an additional button to the gain slider that automatically adjusts the slider to 0 dB output.

Robert