Cubase 5: How to avoid distortion on final mix?

Hi,

This might be a very newbie question, but is there a facility in Cubase to mix down to a stereo track which takes into account the overall level of the total mix?

In Cubase, the sound of the mix seems OK, but when I mix all tracks to stereo and listen in either Cubase or another application it sounds, and is, distorted.

In Cubase, before the mix, there appears to be no indication I’m exceeding 0db.

I know that in the hardware world one would have to drop the levels of all the contributing tracks. Do I really have to use the mixer to pull back each track’s level, by trial and error, until I get a clip free mix?

I’m using an SY1000XG soundcard which has the ability to record the MIDI directly and Cubase Essentials 5.


Thanks.

Rick.

If your master output metering is showing no clipping then you have a different problem?

If your master is exceeding 0dB then you can just pull the master fader back till it stops clipping, not elegant but is fine.

Make sure you mixdown to the correct bitdepth for your intended target. some players don’t like 32bit fp.

Thanks Split, dropping the level on the master fader in the mixer works. I don’t think it’s inelegant!

However, the levels indicated by the output channel meter don’t indicate that the mix will clip. Is this because Cubase doesn’t take MIDI tracks into account in the output metering on the stereo out?

Maybe my setup with an SW1000XG is unusual?

Thanks again.

Ah… well what are you using to sum the final mix?

You could always record the SW1000XG midi output/s into cubase audio tracks.

I would call dropping the master fader inelegant as far as I try to mix internally to a fixed (0dB) master fader and leave a bit of headroom. Probably a throwback to analogue mixing :laughing:

Just listened and checked more closely, the stereo out fader only effects the audio tracks.

So this brings me back to the original question. It appears that I have to adjust the level of eack track, (audio and MIDI) down equally to get an undistorted, accurate mix.

Do other Cubase users have to record each MIDI track to an audio track individually before mixdown? Or are others able to adjust the level of audio tracks (by db) and MIDI (by volume) sucessfully?

Split: I’m doing the final mix via my SW1000XG, this has a second stereo input which can mix all the MIDI and audio tracks to a new stereo audio track.

In analog world, I’d record all the tracks at optimum pre-clip level and then adjust track outputs to accomodate the overall volume level for a stereo mixdown. Where I’m having difficulty with Cubase is that there’s no indication of total level, there’s seperate meters for the MIDI and audio, plus, all sounds OK. Yet when mixed down it’s obvious I’m +3db over.

It depends on the MIDI device. You seem to hear the SW1000XG sounds directly through windows, without them being routed through Cubase thus not included in the mixdown (/just like MS wavetable synth for example). Can you create any return channels for the card´s internal sound channels in Cubase? Probably not… So you´ll have to record them as audio in Cubase first to include them in the audio mixdown.
P.S:You know that the Cubase meter characteristics can be switched between “meter Input” “post panner” “post fader”. So make sure you´re using the correct setting.

Most would use Vsti’s that run within Cubase so that all the audio is contained and routed through the master out.

As suggested it may be worth recording each SW1000sg’s instruments back into cubase individually so you can get better control over their level.

Check what your meters are set to re: pre/post fader. Especially on the main outs.

Split : Thanks for your observation with respect to recording the SW1000XG tracks individually back into Cubase. However this seems like a pretty major step when all I want to do is mix down tracks which already appear in the mixer on individually controllable channels. I will considered it though, and on songs in a much earlier version of Cubase, this is exactly what I had to do.

From what you’ve said it appears that most users are working entirely with audio tracks by the time of a final mixdown, and so would have control and correct level indication of all tracks via the Stereo Out (Master Channel).

I’ll work with your suggestion of using the Stereo Out to control the audio levels and then link all the MIDI tracks and adjust for relative parity: This will still leave a little trial and error as the level indicator on the Stereo Out will only reflect the audio channels.

thinkingcap : Thanks for your thoughts. However the SW1000XG indeed does have it’s own internal loopback, so MIDI tracks are recorded.

mashedmitten : not sure what you mean in the context of CE5; I don’t think I have control of the pre or post fade state. In any case, the individual channel levels are displayed correctly and respond to changes in fader position.

Hi,

Have you tried using Limiter in the Stereo Out with 0 db ?

I think Split’s idea of mixing down all tracks to audio before the final mix is a major part of the answer, not such a problem as of the 31 tracks in the song I’m using as a test only 6 are MIDI.

What I did find was that my soundcard, an SW1000XG, appears to record hot by 3db (from system boot, no previous activity), on the Stereo Input #2 which is the loopback input. By inserting an XG reset at the beginning of the song, this is removed…to oddly a -0.6bd gain. ( I recorded Odb and -6db sine waves at 440 Hz and then used the Stereo Input #2 to record them and check levels.)

Thanks again for your input, I know it’s hard to interpret others technical descriptions and I appreciate the effort.

PKP, I have sucessfully mixed down my test track, but the level is a little low due to only a few high level excursions, I’ll give limiting on the Stereo Out a try, thanks.

PKP: Limiting on the Stereo Out appears to be Post Mix. So this doesn’t help with transients on the way in. I think for this to work I’d have to limit each contributing channel.

Well, obviuously you have a different understanding of “distortion on final mix” and “limiting the stereo out”. The input on the stereo bus is 32 Bit float, therefore won´t clip (unless you use fixed point FX in an insert slot). In that case, turning down the master fader of course won´t help either, and you´ll have to turn down the individual channels…
Only the output (post fader) will clip, when mixing down to fixed point format. Therfore post fader limiting makes sense.

thingcap : Thanks for the post. I think we probably mean exactly the same things by “distortion on final mix” and “limiting the stereo out”, but remember, I’m mixing down both audio tracks and MIDI. The audio tracks respond to the Stereo Out fader and limiting, but not the MIDI tracks.

It’s probably taken a little time for me to come to grips with this as when I monitor the song I’m using as a ‘test bed’, there is no apparent distortion; I have no idea why. (Why would there be an audio path to the line out of my SW1000XG which can accomodate more than 16 bit data?)

I think split has nailed it: one needs to convert all the tracks to audio before final mixdown to get a proper view of levels. This is just hard to understand given the previous lack of distortion while monitoring. It’s also somewhat irritating, as my ‘test bed’ song has 37 tracks. So it’s taken all my free Sunday hours to mix down a 3 minute song which sounded just right on Saturday!

Hi,

You might want to check the control room mixer (devices menu at top, control room mixer) and ensure that the outputs there aren’t over 0.00 db.

Next…

If I have this right (apologies if I’m wrong)

In a very loose manner of speaking, when your mixing multiple tracks, you kind of fill a bucket with sound.
Your Master output is a bucket of sound, but it’s being filled with multiple full buckets from your multitracking…

So, you tip a bucket of sound into the master output that you’ve got close to 0db. You keep adding full buckets, well, the master will overflow - not in a simple addition scale (1+1=2), but on a logarithmic or other inflated scale.

Solution is to mix your tracks to taste.
Play the song to the end to work out the maximum effective output level of the master channel, and take the master volume down so that the effective output level is just under -0db

If you send it off to be mastered, you will want to take the effective output to -3db (you want some headroom left for mixing)

If you want to attempt to master it yourself…

Set the effective output level to -3db
Mixdown to wave
Open up a new project (or a master template)
Import the wave
Add a Multiband compressor on the track, fiddle with it (or use presets) to get your best sound.
If you had a mono mixdown (personally, I track in mono), add a stereo enhancer - fiddle with it to taste.
Insert the 32 band eq - (or eq of preference) and check the different mastering EQ’s. (mastering typically boosts lows and highs to make up for the difference between how speakers pump out sound at low and high volumes)
If your that way inclined, add a maximiser. The steinberg one does fine without presets, but there might be a particular preset which makes your song even better.
Add dithering to the master output channel
Mixdown your mastered track.

Watch your gainstaging. i.e. make sure that the last plugin on your master track is under 0db when you’ve played the song through, if you’ve added the maximiser, make sure that the plugin before it is under 0db.
The maximiser output set to 0db will set the volume coming out of it at 0db (after fiddling with phasing or harmonics or doing some such magic). This is fine if it’s the only active track, if for some reason you try to multitrack your master, you will have the same problem of filling a bucket with multiple buckets as at the start of this reply… Keep your master to one (active) track, do all of your mixing in your mixing project.


All the best with your tunes…

On your mixing project, if you have any plugins on the output channel you will have some gainstaging to account for. (simple, don’t do it - if your EQing the output, what your probably trying to do is master, so leave the mastering element as a seperate project).

Thanks for the tips danbelina. Fortunately my knowledge of audio is reasonably extensive, otherwise I would have been even more puzzled by the results I was getting. It’s quite interesting to see how someone else approaches the task.

Fortunately, with the help of the previous posters, I’ve been able to understand how the problem arose and arrive at solutions, it’s made me investigate the workings a little more thoroughly.

One minor correction :

The overflow aspect is certainly true, but the summing of the audio is strictly linear, it’s just expressed on a log scale of decibels.