Screw the meters hitting red...I want it my way

[keyman_sam]

So I have completed my song and it sounds great and all but it's hitting red on my Cubase stereo out input.

Note, I don't care - I like the sound so it can hit any color it wants. BUT soundcloud (and perhaps YouTube?) will compress the sound if it doesn't hit -1 dBFS.

So, the question is - how do I preserve this sound without hitting RED? I thought I can reduce the input gain of Cubase stereo out channel, insert Ozone Maximizer set it to -1 dBFS and call it a day. I don't want any compression (hate the "pumping effect") as such, I just want the clipping to go away. Is that even possible? Or is that a dumb question?

[nix808]

I think-
if you want it to sound exactly the same as it does with the clip, but not overload on playback,
hard clip it at .999 or 1.
It's not a limiter or a compressor you want, just a hard clip. This makes the signal's maximum <1, and min >-1.
Another option is to render, and then in an audio editor attenuate it. It will now sound the same as your mix, but will not clip.

edit- A hard clip is a distortion though, and the correct thing to do here, if you don't want a clipped signal is just to lower the master. As you say, any limiter or compressor will alter your mix.

[shroom81]

You should never touch the master fader while mixing instead lower all the channels until your master is not in the red anymore. It sounds like you want to keep the sound of your mix when it's clipping so I'd put a distortion plugin on the master and use that instead of the master channel in the red, which is a big no no.

Also leave the mastering until you've rendered your mix out, open that in a new project preferably the next day and apply some light compression/limiting/eq if needed.

[keyman_sam]

It is debatable whether one should touch a master fader or not. In Cubase the internal mixing engine is 64-bit so it doesn't show clipping till the final stereo out, and I bring down the gain so that master fader doesn't clip when rendered.

The only way around this as I see it is to use as transparent a limiter as possible. Or give mastering to a third party...

[simon.a.billington]

Best thing I could suggest is to let it clip, bounce it, then with an external editor lower the volume until it comes in just under -1dBFS.

Or you could find a very brutal, hard clipping limiter, or compressor and effectively do the same thing with that. Lower its threshold to -1 and lower the mix by 1dB so it's hitting it the same way.

You never know your luck, you could even find a better sound this way.

Not sure if any hard clipping limiters though. Logic's compressor has a hard clip setting which may work if you have access to that. I think it's limiter has a "No Over" option which should also just clip it. IK's Brickwall Limiter I think has a hard mode as well. You could also try Waves Pusher, designed for hard hitting dance sounds.

[keyman_sam]

I don't WANT a clipped sound - it's just that I don't want to get punished by digital distribution (iTunes will fail).

To lower the level as suggested I might as well do it in Cubase. I guess so far the only viable solution proposed is to get a soft clipper (like T-racks clipper which I almost bought).

The other solution I can think of is to put the clipper in only those sounds which actually clip so the artifacts are localized. Or a compressor with an internal brickwall like Waves rcomp or REQ.

[simon.a.billington]

Running it through a hard clipper before the master is the same thing as digital clipping. It gives you "No Overs" supposedly. This is assuming that hard digital clipping is what you are getting from your master.

The main difference is you can insert the clipper into you plugin and control where it clips.

So if you want it to sound the same you insert the clipper, lower that threshold until its at a "safe level", then lower your mix by the same amount so it clips at the same place. Or you can just knock back the input level of your clipper to the same level you set your threshold

[keyman_sam]

Running it through a hard clipper before the master is the same thing as digital clipping. It gives you "No Overs" supposedly. This is assuming that hard digital clipping is what you are getting from your master.

The main difference is you can insert the clipper into you plugin and control where it clips.

So if you want it to sound the same you insert the clipper, lower that threshold until its at a "safe level", then lower your mix by the same amount so it clips at the same place. Or you can just knock back the input level of your clipper to the same level you set your threshold

Ah got it. So instead of having it clipped at 0dB I set it to -1 dB and set input gain to -1 dB so that it sounds the same and hits green (minus the 1dB volume drop). Makes sense.

Any good clippers? I'm taking a look at T-Racks. Maybe I'll check out something in ReaPlugs.

[MogwaiBoy]

Take note that of the magic of 32-bit floating point wavefiles.

If you export at 32-bit FP, it accounts for (plots) values above 0dbFS without clipping (as such).

So I would say... leave the mix as you like it. Export it at 32-bit floating point. Open it in Audacity, RX or some other post-editing program that handles 32-bit FP - use an analyzer to determine the maximum true peak, and reduce the amplitude by the necessary amount.

This saves a lot of time...

- Export floating point file of finished, mastered song
- Let's say it true peaks at -0.1db
- Reduce volume by -0.9db
- Now the True Peak is -1dbFS and you haven't altered anything in the mix
- Export as "Soundcloud Version"

This way you can reduce volume and export iTunes, Soundcloud, Youtube, CD, MP3-conversion suitable versions all from the same 32-bit floating point master file, without having to go back to the DAW and mess with limiter ceilings and all that.

[simon.a.billington]

Take note that of the magic of 32-bit floating point wavefiles.

If you export at 32-bit FP, it accounts for (plots) values above 0dbFS without clipping (as such)...

Then there is that.

But if that's the case it should sound the same if you just lowered the volume and it's only the "psychoacoustic effect of a track being louder that what makes it sound perceivably better.

We all get tricked all the time, so there's no shame in it at all. But it is something the OP should keep in mind.

[keyman_sam]

Well as long as I'm inside Cubase I'm in floating point world, so exporting doesn't get me anything. I can see the True peak, etc. in the loudness meter and reduce volume accordingly. The thing is, my peaks can go 5dB-6dB even though the RMS isn't that crazy, so I need to really pull down the fader just for the momentary peaks.

[MogwaiBoy]

I re-read your original post and.. if you do what Simon and I advise, you should be fine. Export at 32-bit floating and then simply reduce the volume by the desired amount after export. Re-export at the new volume and submit that file.

Soundcloud won't "compress" it in the sense I think you're meaning.. Unless I'm mistaken, their algorithm will simply turn it down automatically too. Some services will actually put your track through their own limiter but this is when the track is too quiet and they want to boost the volume for streaming with other content.

iTunes will be more fussy with the -1db thing as they aim for AAX conversions with no clipping overs. Soundcloud however is pretty casual and foolproof - it will try to work with whatever you throw at it.

[xalama qo]

Interesting topic guys.
I have a question: If you export the 32bit FP master file without making any adjustments at all, then bring it into an audio editor, can't you just normalize the audio to bring the peaks below 0db? I mean, is using an analyzer to spot peaks really necessary?

[xalama qo]

Those 5-6db peaks are a bugger though. I'm no expert, but I'd try to clip the individual tracks that are causing those peaks. Often those rogue peaks are the result of 3 or 4 things happening at once, and to my mind it makes sense to try and tame the guilty elements and approach the problem from the ground up rather than trying to remedy the cumulative effects at the end of the chain.
Obviously if hard clipping on individual tracks changes the mix audibly then it's a bad idea, but I'd clip per track that needs it and then also on the master again if needed to catch peaks which may now only be 2-3db over.

Please correct me if that's disgustingly bad practice

[3ee]

Try Kazrog KClip