Recording and Mixing Levels in Tracktion

[21tones]

Thanks for your responses. I think it was the webpage that Tony H gave a link to that got me confused about this in the first place!
I'm reassured by Valley's comment that if it hasnt't clipped on the way in, and it isn't clipped on the way out you can do what you like with it unless you use a badly programmed plug. From my very limited experience I was surprised to find that I could give big eq boosts that make the track meters go well into the red without any obvious problems resulting.
Even more reassuring is Valley's comment that once cleanly recorded into the digital domain there are no issues regarding headroom for post process mastering.
I had a horrible feeling that everything I'd recorded to date would turn out to be unusable later!
However, there does seem disagreement about the quality issue with some commenting that recording at around -15dfs, or -18dfs the quality will be better because of greater dynamic range.
If I do future recordings where the input level is at -18 rather than my previous habits of as high a level as possible without clipping, I'll presumably be able to balance overall song levels at something pretty similar via mixing and mastering activity won't I?

Sorry to ask what probably seem naive questions but I really don't want to waste a lot of work by not currently understanding later stages in the process that I'm currently unfamiliar with.

What levels do T users record at? I'm intrigued given that there is variation among the people who have responded to this post?
Thanks

[tonyh]

Not sure what you are implying by saying you can "give big eq boosts that make the track meters go well into the red without any obvious problems resulting". You can clip your internal sound even if you have recorded at -18dbfs. Also, from what I am told by those more expert than myself, sound degradation can happen before you clip. Best to keep out of the red IMO. As for me, I am trying to stick with an average (RMS) level of near -18dbfs. Peaks will go higher of course but if I can deliver a mix at those levels I leave lots of headroom for the mastering process, which will need some. Now if you are talking about pushing the music as loud as possible to join the "loudness war" then that's another story. I personally want to keep as much of the dynamic variation in my music as possible, so I am not interested in pushing the sound till the final product is just under '0' or even clipping occasionally.

Take care,
A

[jupiter8]

Not sure what you are implying by saying you can "give big eq boosts that make the track meters go well into the red without any obvious problems resulting". You can clip your internal sound even if you have recorded at -18dbfs. Also, from what I am told by those more expert than myself, sound degradation can happen before you clip. Best to keep out of the red IMO. As for me, I am trying to stick with an average (RMS) level of near -18dbfs. Peaks will go higher of course but if I can deliver a mix at those levels I leave lots of headroom for the mastering process, which will need some. Now if you are talking about pushing the music as loud as possible to join the "loudness war" then that's another story. I personally want to keep as much of the dynamic variation in my music as possible, so I am not interested in pushing the sound till the final product is just under '0' or even clipping occasionally.

Take care,
A

Sound degradation does not occur if you go into the red. There's simply no reason why that should happen in floating point. And it does'nt get worse the closer to red you get. There's no reason for that either. As long as you don't clip the outputs your safe. Good practice to keep under clipping though as some plugins act funny when given out of range signals.

And my number 2 pet peeve. Save some headroom for the master engineer. He does'nt have a volume control on his mastering program ?

[21tones]

Tony
thanks for your response. The example I gave was where I had recorded something to peak as close to 0dfs as possible. I then added an eq boost of nearly 12 db and couldn't hear any distortion or clipping. This isn't a practice I would normally carry out but I just wanted to see what happened. It does seem consistent with Valley's comment that you can do what you like once a clean signal has got through the A/D converters.
I don't like the contemporary fashion for overcompressed music, I do prefer dynamic range, and I've never used a compressor on anything I;ve done. That's more inexperience than ideological objection, but I haven't yet had a need to do so.
I've started looking at the RMS version of Tracktion's level meters. Is there a way of getting a single average for a song or song section?

thanks

[ivegotmono]

And my number 2 pet peeve. Save some headroom for the master engineer. He does'nt have a volume control on his mastering program

Usually when people say this they are talking about compression not level. Level can be reduced but compression can't.

[tonyh]

21...like I said, I'm not an expert...still lots to learn, but trying to listen to those who seem to know what they are talking about. Let's see what the other guys have to say re average, but I don't think there is a hard-and-fast answer.
Jupiter...I guess I'm not up to speed on exactly what being in the 'red' involves when it comes to different DAW's so generally avoid it if possible.
Signing off,
A

[sonicfire]

And my number 2 pet peeve. Save some headroom for the master engineer. He does'nt have a volume control on his mastering program

Usually when people say this they are talking about compression not level. Level can be reduced but compression can't.

i always find this funny! whats the deal about normalizing a track to about 80% ?

[SuperFly76]

And my number 2 pet peeve. Save some headroom for the master engineer. He does'nt have a volume control on his mastering program

Usually when people say this they are talking about compression not level. Level can be reduced but compression can't.

i always find this funny! whats the deal about normalizing a track to about 80% ?

Compressing the signal is the issue. It has nothing to do with normalizing or turning the levels up. It has to do with dynamic processing which actually changes the source signal. This cannot be undone by turning down the volume.

[Mr. Tingle]

21...like I said, I'm not an expert...still lots to learn, but trying to listen to those who seem to know what they are talking about. Let's see what the other guys have to say re average, but I don't think there is a hard-and-fast answer.
Jupiter...I guess I'm not up to speed on exactly what being in the 'red' involves when it comes to different DAW's so generally avoid it if possible.
Signing off,
A

I would suggest reading and re-reading Valley's post, it hits the nail on the head when thinking about recording in the digital domain. Red is not red in 32 bit float hosts. In fact, a lot of people around here actually delete the track level meters, as they mean nothing (apart from showing you have a signal (which is handy)). If you come from analogue it takes a while to get your head around. It did me at any rate.

[21tones]

Thanks to everyone for their input to this. There do seem to be different views on this. My key concern when posting this was to try and establish whether my recording practices to date were going to cause trouble further down the line. I'm reassured that a number of people, more experienced than me, seem to think not.
I'll try recording in a different way than I have done previously and see if my ears can identify differences/improvements. I guess that's ultimately what its all about.
Thanks

[alantak]

Mackie documentation is a bit confusing on this matter:

"The LEVEL METER plug-in shows the level of audio being played back from that track. Overloads in the output are displayed in red, in which case the level/pan plug-in (above) should be used to bring that track's level down."

Why?

Anyway, should I use AVG or PEAK setting for the master level meter? What are good situations for each?

I generally ignore individual track level meters, unless I need visual confirmation of relative levels between certain tracks (e.g. "am I crazy is the snare WAY louder than the kick?") But I am careful to keep the master output level meter our of the red, and do this by simply throwing a level filter in front of it and bring it way down.

It sounds like I will be OK doing this.

And my experience has def. shown that clipping, once material is recorded properly, is always in the plugins.

[rpm]

This is all kind of confusing. I managed to record an album that turned out fine without paying any attention to my levels (as far as I remember) with T2. Now, I'm off to record a second album with T5. This is what I found on the T4 quick start guide, so this is what I'll try to follow:

"Tip: With modern 24 bit input devices, setting the input levels such that peaks in the signal register around -5dB on the input meter, gives a good trade off between audio quality, and headroom."

[klangbastler]

i think, there are two parts to look at:
one is the recording, which brings the converter/interface into consideration.
it is not good advice to leave just big headroom, when recording in 24 bit (i assume this as standard for recording, for my points). every converter has, based on the analog part of its circuitry, a sweet spot, where one should aim at, to use its capabilities best.
this will be somewhere around - 20 db RMS! and differs for different models.
when you record to hot (but still under 0 dbfs, so the converter will not clip), you will get distortion from the analog part of your interface (some may sound good, when pushed, but that depends on your taste and how the circuitry reacts to different frequencies/levels).

when recording too low, you will not use the digital converter to its best abilities. low level parts of your signal will have increased quantization distortion and will be converted with less "depth", so to speak.
so this needs to be balanced, to get best results in the recording/conversion process into the digital realm. i would suggest to compare different recordinglevels for one source, bring them to equal loudness after recording and listen to it.
for recording you need a meter, that shows you both, rms levels to target your analog part and peaklevels, to not clip your converter.

second part:
once you have that figured and everything is safely recorded, many of the advices from above are good guidance. leave enough headroom while mixing, this is the "peak at -9 dbfs and look for an overall level of- 24 db RMS" advice often stated, which involves frequencybalancing and use of limiting/compression.

when the mix is ready and it enters the mastering stage, you will have enough "space" to make the last balancing- and loudness-adjustements and finish the audiofiles to a loudness and balance, that they can safely enter into the reconversion to analog with the da converter on the listener side.

both bridges, the bridge into digital and the bridge out of digital are critical stages, where the audioquality can be harmed (i assume decent music-playing/mixing/mastering ).
for the first bridge i need to watch out while using microphones, using converters/interfaces, setting recording levels, etc..
for the second bridge i need to look mainly while mastering, but the mixing process is already the way to this stage.

both involve looking for levels and loudness, but in a different way.

my two cents , happy recording/mixing,

mathias

[rpm]

That's very interesting, I had never realized that one's interface, converters might have a sweet spot. I will try that.

One of the confusing things for me is the different scales we are talking about. What are RMS, dBFS, dbFU, VU, etc.? But more critically, what is the scale that is shown on tracktion's meters.

Do I need an extra VST plugin to take these measurements or can I do everything using Tracktion's single meter?

[klangbastler]

tracktions meters show peak values by default, useful to avoid clipping of the ad converter, which you want to avoid under all circumstances.

when you rightclick on the meter area, you can switch it to RMS.
use that on the inputmeter in the input-properties-panel, to target your - 20 db input RMS level. with most sources, there shouldn't be any peaks over 0 dbfs (switch back to peak to check) with that levelsetting.
with percussive sounds and drums, the peak to average (RMS) ratio may be higher then 20 db, which means, you will get peaks higher than 0 dbfs, when aiming for -20 db RMS.
then you have to lower the input on your hardware and go for maybe -30 db RMS to avoid clipping.

regarding measuring:
all loudness-measures need to have a "timing" window, as they describe changes over time and a reference point (measuring always needs a reference, otherwise you can't measure anything). the confusion comes with throwing them all together and not differenciating their fields of use and what they can show you regarding music and loudness.

you have to look at analog and digital as two different worlds, with different rules.
but in the conversion from ... to ..., they need to match in the right way, to give you best results.

peak values are instantaneous measures (shortest possible timing window) of loudness and they do not give a representation of the loudness we perceive, when hearing a sound.
they just show the loudest little moments that happen somewhere in time.
peakmeters are good for keeping ad/da converters from clipping, as they have a "hard wall", anything over 0 db will give nasty distortion even short peaks (not always true, there are designs, that try to give you a kind of more pleasant distortion when going over 0 db, to give you some safety headroom).
in the digital world, the shortest represented value is one sample, so digital peakmeters show you "sample" values.

RMS values show an integration over a longer timespan, that represents a little better how we perceive loudness. those are typically lower than peaks, by different amounts, which is defined by the sound you measure.
steady tones can have equal peak/RMS values, while percussive sounds will have greater differences up to about 30 db. when you take the 0 db from the digital dbfs (decibel full scale) as reference, you will get negative values for RMS. this the -20 db i was talking about.

i leave it at that for the moment, because english is not my first language and my head gets hot when trying to explain these things as clear as i can ...
feel free to ask.

mathias