Do you know how to read your meters? -18dBfs theory?

[Stromkraft]

That is what the drive knob is for, think of it as a Volume knob.

You're missing the wood for the trees. This is a general topic and not about the Saturation audio effect. Not all audio effects have volume in knobs. Is it even convenient to lower the volume in case there is one?

[Stromkraft]

there's lots of headroom in floatingpoint right until you hit the output stage at THAT point it becomes a fixed point file, at THAT point you should make sure everything is below zero.

That's true, but this point also occurs when recording another channel as 24bit audio. It's not just on the master where you need to observe levels. I recorded a bass sine from a track clearly in the red in a new simple experiment. note that Images are "volume normalized" in order to show the actual waveform better.

Here is a bass sine wave recorded at -12dB (Volume increased for image):


Here is a bass sine wave recorded at 1.3dB (Volume decreased for image):


That's clipping.

There is no magic. It's all applied maths and the option to make informed decisions as a producer. I'd think you'd agree with me on that point.

Thank you for your inspiring observations and experimental ideas. I appreciate them.

[Machinesworking]

That's a great point Stormcraft, I'm curious as to whether the track you were according into was in the red at 1.2db though?
Seems to me it would have to be for the clipping to be there. I never even thought personally that recording to a different track would magically making clipping go away, but I suppose a lot of people would.

Just a side note, but the main thing I think makes sense is to send to the master fader at least a -9dbfs signal so you can apply last minute EQ, compression, and Limiting to the track. It's much harder to master a track when it's near 0db. It's all not as big of a deal for those of you who do everything in the box, but live drums and other heavily dynamic sources like vocals require a lot of babying to get levels to sit in a mix.

[fishmonkey]

individual tracks in Live have higher internal processing headroom.

whenever you record a signal, or send it through a DAC for playback, you do not want to go over 0 dBFS.

[phaded]

Has much less to do with clipping than it does with calibrating a consistent signal throughout your projects. This is both to feed devices along the signal chain a reasonable amount of headroom to do its work as well as making it easier to a/b what you are doing to the signal with any particular processor. But its important to note that it is the *plugins* headroom that you are concerned with and not Live. Usually if a plugin needs a specific level for its sweet spot it will be indicated somewhere in the manual.

What I do with my workflow is drop a VU meter *first* thing on my track chain. I love VUMT by Klanghelm for general purpose VU metering but Hornet Audio's VU meter has a handy auto-gainstage button you can use to get your levels in a pinch. Hornets Track Utility will do the same (autogain ON, set to -18 db RMS). After you have this set maintain the levels with metering in between plugs. I can't stress how helpful this is to understanding *what* you are doing to your sound too. By lowering your output after processing a sound where you use heavy gain you avoid being fooled by things sounding 'better' when in reality its just more loud.

Probably already mentioned but when we talk VU we are talking average level... So a sub tone at 60hz will find 0 VU at a lower peak than say, hi freq percussive sounds (which actually can catch clips / overs at or near 0VU). This is why your peak meters in Ableton can be misleading- its peak energy, not overall loudness. Hope this helps. Hard to get your head around the concepts and especially the 'why?' in the digital domain, but I can attest that its helped my work in the studio tenfold.

[Stromkraft]

That's a great point Stormcraft, I'm curious as to whether the track you were according into was in the red at 1.2db though?

No, only the source track was overdriven as in Angstrom's example.

It's much harder to master a track when it's near 0db. It's all not as big of a deal for those of you who do everything in the box, but live drums and other heavily dynamic sources like vocals require a lot of babying to get levels to sit in a mix.

Yeah, this is the one point I've taken to heart. If you keep a low peak level and a reasonable RMS level you can add additional volume later, as can the master engineer. If you're already up there nearing 0dB everything must come down first which is simply too much work for very few advantages IMHO. Especially if you do it for experiencing volume, in which case you can just raise the volume of your monitors to the level that suits you.

[Stromkraft]

Has much less to do with clipping than it does with calibrating a consistent signal throughout your projects.

Are you arguing now the displayed signal isn't clipped? It is clipped because it was recorded with a too high input volume as the source track was driven over 0dB. I wasn't expecting this result though and I welcome other's testing the same. I just decided to learn more by experimentation.

This is both to feed devices along the signal chain a reasonable amount of headroom to do its work as well as making it easier to a/b what you are doing to the signal with any particular processor. But its important to note that it is the *plugins* headroom that you are concerned with and not Live.

Yes and no. Yes, I set up the proper levels for my plug-ins as well as a reasonably consistent volume level trough the whole signal chain for one sound source. Especially when using plug-ins to avoid the "louder-is-better" syndrome that many people suffer from (some I work with too unfortunately).

No, because as I just demonstrated when making a recording of another track, which is something people do, it is indeed how Live handles the signal you must be aware of.

What I do with my workflow is drop a VU meter *first* thing on my track chain.

Yes, I use the Free-G from Sonalksis. It isn't a VU, but shows RMS levels as well as peaks. It doesn't show intersample peaks, so isn't really suitable for mastering I think, though I use it last on the master output.

[Stromkraft]

To further illustrate the effects of Live's internal floating point calculations that Angstrom was talking about previously as well as 32bit internal resolution I made a variation of the previous little experiment by simply overdriving the track with 15dB using Utility and then freeze/flatten. The results as follows.

This is the recorded sine unzoomed:


Sounds like it looks.

This is the recorded sine zoomed out so we can see the whole waveform:


No clipping that I can see or hear.

[Stromkraft]

recorded a bass sine from a track clearly in the red in a new simple experiment. note that Images are "volume normalized" in order to show the actual waveform better.

I have now repeated the same recording procedure but this time using resampling with the Master set at 0dB which gives a clipped signal which translates into recording.

Indeed if you lower the volume level of the master so you're in the green the recorded waveform is not clipped.

My point is still that if you never operate too close to 0dB somewhere in your signal chains as well as never record red hot signals in the first place — It's a pretty stupid thing to do —  these are not things you need to worry about.

[phaded]

Has much less to do with clipping than it does with calibrating a consistent signal throughout your projects.

Are you arguing now the displayed signal isn't clipped? It is clipped because it was recorded with a too high input volume as the source tracks was driven over 0dB. I wasn't expecting this result though and I welcome other's testing the same. I just decided to learn more by experimentation.
.

I think you misunderstood me, but then I suppose I didn't start off entirely clear. That would be the pitfalls of trying to write something technical on a smartphone for me.

What I am saying is: *the reasoning* for calibrating your levels to average around 0db VU has very little to do with clipping. The calibration is more aimed at signal consistancy and to allow proper levels to be fed into plugins that operate with a limited amount of internal headroom.

In other words, the plugins in your signal chain *could* be working with less headroom than you would normally expect in a DAW. So in order to avoid unnecessary distortions and get the best out of them, you will want to feed them an appropriate signal level

There is nothing I'm really challenging as far as the math is concerned... nor am I referring to any picture of a clipped signal. Because in the end 0db VU is so far under fullscale that you really should never be near clipping if you're gainstaged right =)

So.. in a nutshell

0db VU = AVERAGE loudness (watch how the ballistics of the meters are slower). 0db VU is typicially calibrated to one of the following: -18db RMS (most common), -14 RMS (hotter), or -12RMS (pretty smoky). Best way to find out if this applies to a particular plugin is to consult the manual as I mentioned before.

0db peak = what you see on Ableton's meters, giving you the loudest peak of the signal. If the RMS part is confusing its easy enough to keep things around -10db peak (or less) and things should be groovy.

[Stromkraft]

What I am saying is: *the reasoning* for calibrating your levels to average around 0db VU has very little to do with clipping…

Yes, I agree with the gist of your entire post. That's what I wanted to get across. There are very few good reasons to not control your gain structure. It's up to every producer what levels to maintain that they want to work with of course.

[phaded]

What I am saying is: *the reasoning* for calibrating your levels to average around 0db VU has very little to do with clipping…

Yes, I agree with the gist of your entire post.

Can I get one up top?

[Machinesworking]

there's lots of headroom in floatingpoint right until you hit the output stage at THAT point it becomes a fixed point file, at THAT point you should make sure everything is below zero.

That's true, but this point also occurs when recording another channel as 24bit audio. It's not just on the master where you need to observe levels. I recorded a bass sine from a track clearly in the red in a new simple experiment. note that Images are "volume normalized" in order to show the actual waveform better.
That's clipping.

OK I did your test. The main thing that affects the ability of Live's tracks to record without distortion is whether it's Pre FX or Post Mixer. If it's Post Mixer then you can overload the track, but Pre FX the audio does not overload and distort. Pre FX is using the floating point headroom and post mixer is summing? dunno for sure but....

Anyway, this lends credit to the idea of thinking of the summing/mixing bus as -18dbfs rather than relying on the amazing ability of Lives tracks to not distort. Also that if you do want the added volume of a maxed out track in Live the best way to feed it to another track is pre FX.

[fishmonkey]

surely any time you render the audio the maximum possible recorded level is 0 dBFS before clipping occurs...

[lunabass]

Exactly. That there are devices that react dynamically was the main point I made.

Ahh, so that was the point you were trying to make, you simply sounded like someone that had no idea how Saturator worked. Of course you cant assume that a higher or lower volume is better on a plug-in that reacts dynamically to a varying input level!

This entire thread was about -18dBFS (0VU) being the digital "sweet spot". The argument often used by people to back this up is that hardware emulating plug-ins will benefit from -18dBFS as that's how the hardware operates. Every test I've done simply shows that this is not true. I'd love someone to show me a plug-in where this happens (honestly)

Yes, recording a signal at 1.3dB causes clipping, who was advocating that this would be ok?

To be clear, I'm not saying we should have all levels up at 0dBFS but I'm calling bullshit on -18dBFS as being the sweet spot (unless you are using hardware)

Utimately this:

It's the result you want that should guide you.

Is what has got me to the opinions I hold now, seems we can both agree on that