Reducing # of Audio tracks to decrease noise?

[Phuturistic]

I've been interested in the S/N ratio of my system lately and anything I can do to get the most out of it. For starters, I'm recording MIDI sequences of anywhere from 20 to 25 MIDI tracks to stereo audio tracks in Digital Performer, to be mixed ITB, processed as necessary, and burned to cd. I've recently been looking at noise floor levels, usually at the very beginning of my recorded audio soundbites before the signal comes in. I'm noticing that noise floor levels are hovering around -81 to -83dbfs with a signal peaking around anywhere from -18 to -12dbfs for most tracks I've recorded, which I imagine is pretty good (correct me if I'm wrong?). However, when I record 10 to 15 MIDI tracks to their own stereo audio files and playback those files together in the mix, that noise floor starts to add up and become more noticeable. By soloing two stereo audio tracks instead of just one, and checking the overall level on the master bus, noise jumped from around -83dbfs to -72dbfs.

Obviously the noise isn't going to jump 10dbfs with every track I add to the mix, but the added noise isn't helping my efforts. For the record, I'm recording hardware MIDI sound modules through a Brent Averill 1272 clone into a modified MOTU 828mkII. I might get even better S/N results by removing the 1272 from the chain, but I haven't tried removing that from the equation yet. I'm thinking if I just group MIDI instruments together and record each group as a single stereo track, while bussing different groups to compression, eq, etc. I can manage noise more efficiently that way. Or maybe I just need to cut out dead noise on stereo tracks, such as in between snare hits and kicks, etc. Or (and I know everyone seems to hate this one) I could record a little hotter going into my computer and just bring down levels once I'm mixing everything ITB that's already been recorded to disk.

Any opinions on the topic would be appreciated. Thanks.

 

[Phuturistic]

To add to what I said earlier, I should've looked at a couple other specs. I found out that the dynamic range of the hardware module I'm using, an E-MU Proteus 2000, is >90db, with an S/N of >92db. My 828mkII's A/D conversion has a dynamic range of 107db. There were a couple MIDI tracks that I pushed my 1272 a little more to get the signal peaks to sit around -12dbfs (-12dbfs seems to equal 0 vu on Digital Performer's analog style metering) and that pushed the noise floor to about -75dbfs, which seems to be an acceptable level. I guess my real question is, what noise floor levels are you guys getting on average when tracking to disk? What would be an unacceptable noise floor level?

 

[mshilarious]

I don't do MIDI, but when you're recording audio, especially in a home setup, it's tough to get noise much below 20dB, so if you end up at 70dB dynamic range I think that's a pretty good result. You can help alleviate the noise problem from large track counts by muting tracks where they aren't playing.

 

[gecko zzed]

I do a fair bit of midi, but I rarely record each midi instrument in its own audio track.

That means that when mixing, I am mixing a combination of tracked audio and midi. Once I am happy with the midi mix (even if there is still unfinished business with the audio), I convert that to a single stereo audio track. This is one way of reducing noise generated by having numbers of tracks.

 

[mixsit]

. . I'm thinking if I just group MIDI instruments together and record each group as a single stereo track, while bussing different groups to compression, eq, etc. I can manage noise more efficiently that way.

Do the midi tracks seem to be the worse offenders, and are they're midi levels optimized (prior to bouncing to audio)?
Wayne

 

[SouthSIDE Glen]

You could run each of your tracks through a quality noiseprint-style noise redcution plug.

G.

 

[Phuturistic]

Do the midi tracks seem to be the worse offenders, and are they're midi levels optimized (prior to bouncing to audio)?

That was another issue I considered as well. My general practice is to record any MIDI tracks to audio at about -12dbfs. At first I was turning up the output of the 1272's to assist with reaching that level. But of course, by doing that I was bringing up the noise floor of the pre-amp (either that or the pre-amp was amplifying the noise floor of the Proteus 2000, I'm not sure which way it works). Anyway, I went back in to my MIDI tracks and re-adjusted some of the on-velocities, which brought up the level of the MIDI track(s) and allowed me to back off on the 1272's gain. Mind you, I'm not pushing the 1272 very hard, never over 50% on the output gain control.

I found that pushing the 1272 at about 25% on the output resulted in noise floor levels around -83dbfs, and when I pushed it at about 50% noise floor sat at about -74dbfs. This is all with signal levels peaking around -12dbfs, like I mentioned earlier.

 

[Phuturistic]

You could run each of your tracks through a quality noiseprint-style noise redcution plug.

Hey SSG,

What would you recommend as a good noise reduction plug? Anything in an Audio Unit format?

 

[Agent47]

Do you have to record all your MIDI tracks as audio? What I mean is, in my case I use a lot of MIDI, but I'll just record the whole finished, mixed sequence as a stereo wav. I'll leave MIDI as MIDI until it goes to final stereo wav. Any EQ or compression I can add using the synth's own internal fx.

Obviously you lose the luxury of being able to use separate plug-ins on each part, but it would give you better signal/noise ratio.

 

[ecktronic]

You could run each of your tracks through a quality noiseprint-style noise redcution plug.

G.

Im not sure that is the solution to this problem.
I would sugest getting to the route of the problem rather than using a plug-in to try and solve your problem.
Noise reduction will take some of the good qualities from the tracks.

Eck

 

[SouthSIDE Glen]

Im not sure that is the solution to this problem.
I would sugest getting to the route of the problem rather than using a plug-in to try and solve your problem.
Noise reduction will take some of the good qualities from the tracks.

Eck

I would normally agree with this, as I am from the general schools of "the less processing the better" and "fix the problem at the source". I agree completly with your inclinations here.

However, looking at his signal chain, I'd be hard pressed to blame most of the noise floor he's getting on anything other than the preamp. As he said himself, removing that pre from the chain may help him out, but I'm not sure that I can recommend a replacement of similar character that's not also going to give him similar noise. If he's willing to sacrifice the coloring for the cleanliness, that's OK.

So in leiu of that option, I'd personally go the NR route. A < -80dBFS noiseprint taken by a quality NR product would have extremely minimal effect on the signal, IMHO. YMMV ETC

Phuturistic, as far as a recommendation, AU is not my strong suit as I'm running a VST/DX combo platform myself. The only one I know of myself that'll do AU and that I'd recommend for quality would be the Waves Restoration bundle. It does a great job, but it is admittedly a bit on the pricey side. Maybe someone else knows more about AU options than I and can chime in on that.

G.

 

[mixsit]

.. Anyway, I went back in to my MIDI tracks and re-adjusted some of the on-velocities, which brought up the level of the MIDI track(s) and allowed me to back off on the 1272's gain.

That's where I was going. From what I've seen (very little BTW and with hardware synths) there can be quite a high noise floor if the velocities aren't up.
Wayne

 

[RAMI]

I never understand why people RECORD their MIDI tracks...Isn't the whole point of using sequencers, etc...to not have to print them? I used to do MIDI bass and drums and never recorded them onto tracks.

So what am I missing?

 

[Phuturistic]

I never understand why people RECORD their MIDI tracks...Isn't the whole point of using sequencers, etc...to not have to print them? I used to do MIDI bass and drums and never recorded them onto tracks.

So what am I missing?

The reason MIDI has to be recorded as an audio file is because MIDI is simply performance data. So you can have an entire sequence of several MIDI tracks playing back performance data, but all that's doing is telling whatever soft synth or hardware sound module your using to respond to the MIDI data your sending to it. In the case of soft synths, most times you can just "freeze" the output of the soft synth along with the MIDI tracks that are playing it, to a wav or aiff file.

In the case of hardware sound modules, you have to actually record the outputs of the module to a wav or aiff file, in order to burn the recording to cd, to be mastered, etc. The point is, with a MIDI hardware module, the MIDI data your sending to it is telling it what to play, and the outputs of the module that go into your soundcard/interface are actually capturing the audio that its playing and storing it as an audio file.

 

[Phuturistic]

Obviously you lose the luxury of being able to use separate plug-ins on each part, but it would give you better signal/noise ratio.

And that, I believe, is where the balance lies. I need to be able to seperate certain tracks for certain purposes, such as drums together for compression, strings, pads, and other sounds that I would like to independently apply delay and other effects to. I need some flexibility with processing different tracks seperately, but also, as you said, keep my S/N at optimal levels.