Problems with Levels on SoftSynth audio

[Bruce Olsen]

Background: I've started writing drum and percussion parts in MIDI for our band. I send the resulting audio files to our engineer for mixing/mastering. Basically, I'm acting as my own recording engineer. When I played real drums, I was my own roadie so this is way better.  For the most part I don't add any FX, I just look for decent drum and percussion synths. In a few cases I've used the PX-64 percussion strip (say, for reducing the overtones on a ride cymbal bell).

My objective: to give our engineer dry mono audio multitracks somewhere around -12 dB. so he can do the rest.

My problem: I'm having huge difficulty getting any consistency between the levels I see when I'm

• writing/editing a song
• recording it
• exporting/re-importing before I send the component tracks to him

At the end of the process I'll have some tracks with significant clipping, and also some tracks down around -15 or -20, even though I tweaked them at every step of the way to remain close to -12. Both CW and Audacity show me basically the same thing. I handle all the tracks at the same time (meaning--I believe-- that all my settings should be consistent across tracks) so this inconsistency is really puzzling, so there must be something huge I'm just missing.

My process so has been to arm the audio tracks for recording (not using bounce) and then exporting them (and I'll add an aside here about how much of an improvement the brand new export feature is: thanks to all involved, it's very nice--especially the file name tags). I doubt bounce would make a difference based on the strange level behavior.

I've looked for hours for some info on this, and as thorough as the reference guide is, it's still a reference and if there's anything in there (or in this forum) on this I can't find it.

Can anyone suggest things what might have caused this, as well as the types of things that could go wrong in the process? 

Thanks!

[David Baay]

I would say step one is to set the Volume widget  of the MIDI track (or MIDI side of the Inspector for an Instrument track) so that the VSTi will always be initialized to the same level (CC7 master volume is sent every time you open the project, and every time you start playback). By default the MIDI Volume widget will show (101). The parens indicate that it's disabled. Moving it automatically enables it, and you should see the level control in the VSTi's UI follow its movement. Depending on what synth you're using, and what tracks/channels you have set up it may be the volume of a single channel or the volume of the instrument's main output. If you have more than one MIDI track riving the synth, you should enable the Volume widget on only one track (right-click and Disable the others as necessary). If more than one MIDI track has volume enabled, the level of the highest numbered track will supersede earlier ones.

Beyond that you would need to give us a run down of you track/bus/send routing, what drum plugin you're using and how it's configured, how you're entering MIDI (drum controller, keyboard controller, PRV or Step Sequencer) , and let us know what export options you're using. I generally recommend exporting with source = buses, and unchecking everything but the Master bus. This generally ensures the level of the resulting audio file is the same as what you see in the Master bus meters, and what you hear while playing/recording.

My standard workflow is to bounce the Master bus to a 'Master Bounce' track in the project that routes directly to the main hardware outs in parallel with the Master bus, and group the mute on that track in opposition with the Master bus mute so I can A/B between them. Once I'm confident the rendered bounce is an accurate representation of the live mix, I'll export just that track.

 

 

[Bruce Olsen]

Thanks very much for the ideas, I'll take a look at the things you suggested.

As I continue to think about this it feels like I'm missing some important concept.

Some answers for you:

• The MIDI was entered by hand in the PRV; I don't have any controllers. 
• A Drum Map is used to display meaningful note names and send the MIDI to the plugin's input.
• I'm not assigning MIDI channels, but this is the only MIDI device in the project. (I assume I should start managing MIDI channels in the future as I develop more complex pieces with multiple plugins?)
• I'm using the free Steven Slate SSD5 drum plugin, which is actually a sampler, not a synth. 
• I direct each instrument sample (Kick, Snare Top Head, Snare Bottom Head, etc) to a distinct SSD5 output. I'm interested in 11 of the samples.
• In addition to the MIDI track, my project has 11 audio tracks, each assigned to one of the outputs I'm interested in. I've verified this mapping more than once. 
• Each of those 11 audio tracks is eventually sent to my engineer, so by my understanding I wouldn't be able to use bus outputs.
• I also have about a dozen other reference audio tracks (guitars, say, or a previous full mix from my engineer) that I listen to when writing /editing parts. I don't record them, they're not a problem.

Here's the process I've been following to create the audio in those tracks--when done, my expectation is that each of the 11 audio tracks has all FX baked into it.(Is that correct or a misconception?) 

• The Gain control and the slider on each of the channel strips start zeroed out
• I write/edit in the MIDI track with the PRV. If one of the tracks is too loud I'll adjust it with the gain control, leaving the slider alone. I don't care about the levels, I care only about the way it sounds. This may be changed often as I work on various parts in MIDI.
• When the MIDI parts are ready, I listen through a few times and adjust the gain control so that the peaks are around -12. The overall mix may not sound good, because it's about the levels, not the sound.
• I arm all 11 tracks and hit record (I haven't used bounce--does it make a difference?) 
• I watch the levels as I record, and here's where things start going bad. I've found levels go way up here, or way down, with clipping on the resulting audio. So I tweak the gain control and try again. (I don't think I always delete the bad audio files: does it matter?)
• After I have the levels all around -12  I'll re-record the tracks as above.
• When that's done, I'll mute the MIDI track and listen to my 11 tracks again--and once again, the levels will be all wonky. Some ridiculously high, some very low

As far as routing, here's a rundown:

• 11 audio tracks (recorded as described
• The tracks all route to 2 busses, 1 for all cymbals and 1 for all drums. This isn't realistic for mixing but... I'm not mixing.
• The 2 busses route into a drum master bus.
• The drum master is routed to a bus called Master.
• I also have about a dozen audio tracks from the other band members routed similarly (i. e., there's one bus for the lead guitar, one for the bass, etc. and all of those busses also route into the Master bus.) I use these to gauge what I'm doing on drums.
• The Master bus is then routed to a track called "Master Mix". We have sometimes used BandLab at early stages of demoing, and I sometimes do a quick mix of a demo drum part in CW and upload the drum mix track to BL. The output of this "Master Mix" bus is assigned to my headset (don't @me).

The drastic nature of the level changes makes me wonder if phasing has a role to play in this...?

Anyway, thanks for any light you can shed on this.

Edited September 17, 2021 by Bruce Olsen

[57Gregy]

11 hours ago, Bruce Olsen said:

I arm all 11 tracks and hit record

I don't really understand this.
If there is data in the MIDI tracks, it is already recorded.
To get the audio from the soft synths, you can Freeze the tracks of Bounce the tracks or export the tracks.
Maybe it's something new in CbB that I haven't heard of? Or something old in CbB I've never heard of. ?

[Bruce Olsen]

Thanks for responding.

Did you see anything about my process  that might shed light on the problem with levels changing? Any comments on my questions?

1 hour ago, 57Gregy said:

If there is data in the MIDI tracks, it is already recorded.

There's just MIDI data, not the audio I needed.

I learned about arming tracks first and it seemed perfectly natural to me: play the MIDI and record the audio it creates. 

I avoided freeze because it doesn't handle sidechains (I don't have any at the moment but I'm a fan of picking 1 method that always works instead of having to carry a lot of exceptions around in my head, only to be bitten in the butt by something I forgot.

I'll try bounce, though if it doesn't let me check/manage levels in the process it may not help me debug this issue.

 

 

Edited September 17, 2021 by Bruce Olsen

[57Gregy]

Well, I think freezing or bouncing would be faster.
I don't know why you're getting these results and I've never tried to do it that way.
Read up on Automation.
Once you have your recorded, bounced or frozen audio tracks, you can automate the volumes of the tracks by riding the faders while it's playing and be able to easily adjust any 'wrong' levels you may have entered by moving the nodes you created. Of course, that should be done without riding the faders while recording the tracks, your method.

[Bruce Olsen]

Thanks for the input!