Re: [AH] "calibrating" my audio mixer?

From Andrew Horton
Sent Thu, Dec 27th 2018, 01:58

Even -18 peaks is totally fine when recording at 24 bits. You can record SO m=
uch lower in 24 than 16 without worrying about getting anywhere near the noi=
se floor on even the cheapest gear. I also heartily recommend Katz=E2=80=99 b=
ook - I=E2=80=99m a professional mastering engineer now.=20

Sent from my iPhone

> On Dec 26, 2018, at 8:41 PM, Brian Willoughby <xxxxxx@xxxxxxxxxxxx.xxx> wr=
ote:
>=20
> Ha!
>=20
> Most of the time, Andrew is right. It has been my experience that -18 dBFS=
 is a good target, although I place my peaks there, not sustained sounds. Bu=
t that fact is that this is hardly a standard that anyone can rely upon. Som=
etimes you have to go to great pains to calibrate the meters so that 0 dBFS a=
ctually reflects any sort of known reference level. Bob Katz goes over the s=
teps in his book.
>=20
> At the other end of the spectrum, many analog synths have terrible perform=
ance with the master volume at 12 o=E2=80=99clock. Not all are bad there. In=
 fact, some might work best there. But at least a few synths require maximum=
 volume if you=E2=80=99re going to get decent signal levels out. It=E2=80=99=
s also far more repeatable to aim for max volume than to aim for 12 o-clock,=
 unless there is a detent there (and even then there can be variation). Agai=
n, there is no standard rule that we can rely upon.
>=20
> You really have to get to know your gear.
>=20
> Brian Willoughby
>=20
>=20
>> On Dec 26, 2018, at 5:32 PM, laurie <xxxxxx@xxxxxxxxxx.xxx> wrote:
>> and on planet earth ?
>> :)
>>=20
>>=20
>> ---- On Wed, 26 Dec 2018 19:25:51 -0600 Andrew Horton <xxxxxx.xxxxxx@xxxx=
l.com> wrote ----
>> Set all synth self volumes to maximum for best s/n and gain. Set sustaine=
d/average levels to -18 - Peaks can go higher. Most ADC=E2=80=99s are calibr=
ated so that -18dbfs =3D =E2=80=9Canalog zero=E2=80=9D on a traditional VU M=
eter and are designed to sound and operate best at -18.=20
>>=20
>> Sent from my iPhone=20
>>=20
>>> On Dec 26, 2018, at 7:14 PM, Brian Willoughby <xxxxxx@xxxxxxxxxxxx.xxx> w=
rote:=20
>>>=20
>>> This is a tough question to answer, partly because I think there are two=
 aspects to the answer.=20
>>>=20
>>> 1) Avoid distortion from clipping; maximize signal-to-noise.=20
>>>=20
>>> 2) Postpone adjustments for loudness until the final mixing stage. There=
=E2=80=99s a difference between optimizing per-channel S/N and adjusting the=
 loudness for a good mix. Don=E2=80=99t confuse the two by trying to handle t=
hem at the same time with a single control.=20
>>>=20
>>>=20
>>> On the first topic, I would hope that every synth/drum machine would hav=
e been designed so that it will not clip its own mixing stages. However, we c=
annot rely on this because many audio circuits over the years, even from wel=
l-respected brands, have included design mistakes that are less than optimal=
. Therefore, it seems that you will have to determine whether each of your s=
ynths/drum machines clip internally when the master volume is at maximum, an=
d if so then you=E2=80=99ll have to discover the amount that the volume need=
s to be reduced until there is no distortion (unless, of course, this distor=
tion is part of the sound that you=E2=80=99re trying to record).=20
>>>=20
>>> Part two is whether your =E2=80=9Cmixer" input can handle the maximum vo=
lume of each synth/drum machine. In my experience with the MOTU 896HD, the i=
nput levels can only be boosted, not attenuated, so there were many recordin=
g situations where I needed to reduce the output level on the sending device=
 because it was otherwise too strong for the 896HD to handle. Analog mixers t=
end to have switchable pads, and my Metric Halo Labs interfaces have program=
mable Trim that can reduce excessive input levels. If you don=E2=80=99t have=
 this option, and if any of your synths/drum machines are =E2=80=9Ctoo hot,=E2=
=80=9D then you=E2=80=99ll need to reduce the master volume.=20
>>>=20
>>> Personally, even when I have the option for Padding Line level inputs, I=
 rarely boost a device=E2=80=99s output level if it means that the mixer wil=
l need to be Padded. Those pad resistors can add a little noise, so it=E2=80=
=99s better to reduce the send level so that the Pad can be disengaged. The r=
esult will be cleaner. Of course, it=E2=80=99s generally true that more sign=
al level on a cable is better for S/N, but once you=E2=80=99ve reached the l=
imits of a +4 dBu Line Level input (which is about +26 dBu!), then you alrea=
dy have more than enough S/N. Boosting above this, to the point where you ne=
ed Pad resistors, is not going to improve anything at all.=20
>>>=20
>>>=20
>>> Finally, you don=E2=80=99t want to sacrifice the signal-to-noise perform=
ance of an individual channel by turning it down to adjust its loudness to m=
atch anything else. Psychoacoustics research tells us that there is a differ=
ence between the level of signal as measured by voltages and meters versus t=
he loudness perceived by the human hearing system. Since noise gets worse as=
 you turn things down, I recommend avoiding the temptation to match loudness=
 between synths/drum machines, and just maximize the channel S/N. Once you h=
ave a mixer full of channels that have minimal noise, you can mix based on l=
oudness and retain the option of making some devices louder than others duri=
ng part of a song, while other devices might be louder in other parts of the=
 song.=20
>>>=20
>>> Looking at the above from a different point of view: If you=E2=80=99ve a=
lready maximized S/N for every channel, then you no longer have the option o=
f turning anything up. Thus, balancing loudness means turning some things do=
wn, which just makes the noise worse. Loudness adjustments should be a creat=
ive step, possibly automated, and certainly dynamically changing during the p=
rogression of a piece of music.=20
>>>=20
>>> Another issue to keep in mind here is that the presets on any synth or d=
rum machine will vary drastically from one selection to another. If you try t=
o adjust levels to match loudness across devices, that balance will be lost a=
s soon as you select a different sound or a different bank of sounds on any o=
ne or more synths/drum machines. Instead, if you adjust for maximum signal-t=
o-noise, as described above, then your mixing process will be most successfu=
l at matching (or mismatching!) loudness, as appropriate for the art.=20
>>>=20
>>>=20
>>> Bottom line: These questions are all part of the general topic of =E2=80=
=9Cgain staging=E2=80=9D - which is very important in a recording studio. Th=
ere=E2=80=99s a lot to learn. I recommend Bob Katz=E2=80=99 =E2=80=9CMasteri=
ng Audio: the art and the science=E2=80=9D as one starting point. It won=E2=80=
=99t cover all aspects of recording studio gain staging, but it certainly co=
vers all of the technical basics that you=E2=80=99ll need. If anything, Bob=E2=
=80=99s book is probably more appropriate for your situation, where you have=
 a digital MOTU interface.=20
>>>=20
>>> Brian Willoughby
>=20