One-Synth-Challenge: General discussion thread

[Frostline]

There are a couple reasons you're not supposed to work with rendered audio:

One main issue is that many synths have free-running oscillators or inconsistent sound, if you could render to audio, say you don't like how one particular phrase came out, you could re-render just that one phrase and circumvent the annoying behavior of the synth. I don't think there have been a lot lately that seem to fall into this category, but there have been some in the past. Especially when it comes to things like percussion.

Oh we had that problem this month. The delay. Someone else also had issues with it. Automate it to come on and it might work just fine, freeze the track and the delay would start echoing notes from the previous measure and totally mess up the effect. Unfreeze the track and play it, it might be fine, it might be still messed up. It totally hinged on when play/render was was pressed. I bet I have 30+ track freeze stems solely due to this issue. But instead of re-rendering the problem phrase I had to re-render the entire track.
At the end of the day though, what is the difference? Are we not trying to make the synth sound as good as we can possibly make it? Either way it was accomplished I didn't want the glitchy noise in my track. Should I have left the glitch in since it was what was produced? Should my track be DQed because I did multiple renders to time the effect to my liking?

A secondary issue is reproducibility. If someone checked your track and it was a bunch of waveforms that had been moved around or whatever and no longer match up to the source MIDI or automation or whatever else, it would become practically impossible to determine that the provided project was really where the sounds came from, that the whole track as it stands is something the synth can make...

The argument to this I am going to provide is a bit of a weasel, but....

I see nothing in the rules pertaining to what DAWs are allowed. So really there is no guarantee that the file I might provide the arbitrator as 'proof' of the sounds emanating from a particular VSTi would even be a file format they could open on their computer. Now while I personally could not, I am confident there are a few folks out there who could roll their own OS and DAW on some crazy one-off hardware where the VSTi is running in some type of wrapper. So what is the rule then?

Or maybe a slightly less weaselish argument might be that there might not be a file at all.
Say I get 3 good keyboard players each with a PC running a host for the VSTi, each PC has a control surface where the knobs and sliders have been assigned to parameters and 3 sound designers are running those. A left and right channel from each are fed into a mixer with a digital recorder attached. The recording is then submitted. No file is ever created other than the wave recording.
Depending on the people involved it might make a pretty interesting track. Envisioning a Jarre, Kitaro and O'Hearn jam session or something.
But it doesn't look to be against the rules.

[z.prime]

Well, the latter would not be allowed, as you can't record audio

But the unfortunate reality is, yes, your track probably should be disqualified if you made multiple recordings and picked the best one to work around oddities/bugs/quirks with the synth. This is precisely one of the reasons the rule is in place. Don't shoot the messenger.

[Frostline]

Well, the latter would not be allowed, as you can't record audio

Aww shucks, I already had Jean and Patrick on-board with doing the project next month. Was even gonna ask if you wanted to fly out and be one of the sound designers. Now I'm all bummed out.

But the unfortunate reality is, yes, your track probably should be disqualified if you made multiple recordings and picked the best one to work around oddities/bugs/quirks with the synth. This is precisely one of the reasons the rule is in place. Don't shoot the messenger.

Well, ignorance of the rules is no excuse I guess but I really didn't know taking a second (or twelfth) swing at getting a section to sound right due to a glitch was illegal.

But...
Looking at the situation of how I might respond when a buddy asks "How'd ya do in that music thing?"
My pride likes the sound of "Oh, I got DQed because I tried to cover a synth glitch" over "Oh, dead last."
So I'll take the DQ

[z.prime]

Right, well, the typical example is with drums. Let's say you have some drums that sound good for 3 iterations, or maybe just 1 and then they go flaky, so you sample that one drum and use it in a sampler or just repeat the waveform right in the arrange window. It was something the synth couldn't really do because obviously it had some hits that came out flawed much of the time... this is actually fairly typical for any synths that don't restart the oscillators. So, by sampling the drum sound to get one good one, you've worked around a limitation of the synth. This is explicitly outlawed. Now, expanding that to a short phrase is much the same idea... the synth had a quirk that got worked around by picking the best repeat of it. I don't really know where the line is, which is why 'freezing' is perfectly fine since it's the entire track's worth of content even though, technically, you could have a great freeze render vs. other ones that were worse... It's possible a phrase like 8-16 bars might not be much of a problem as long as the MIDI stays in sync with it, I can't recall ever having a specific hard line on this. But, like other things, it could be a bit gray... many of the allowed FX are sorta easy to overuse anyway. I'm actually all for doing all kinds of crazy stuff with any FX you want and sampling percussion (for a few reasons - including saving CPU), but as soon as you do that, you enter the world that makes it difficult to determine where the sound actually even came from... as for choosing DQ over last, you probably made the right choice

[zerofox97]

Y'know, I've been wondering for a while now. If someone were to make an LA synthesis VST based on say the D-10 or MT-32, would only the drum samples not be allowed or would all of its samples not be allowed. Just asking because it's kinda an interesting subject to me, but alas, there is no such plugin.

[Guenon]

By freezing a track, you can simply unfreeze them all and re-render the track and see what it sounds like, it should be essentially the same...

Would it be okay to use subprojects for this instead of freezing track by track? I didn't think this would be necessary, but I'm asking because I'm experimenting with Tyrell here, and it seems to be remarkably heavy on the CPU when enough instances are triggered at the same time.

What I mean is: rendering a subproject (whole tune's length in one go) and loading the resulting WAV on an audio track, doing absolutely zero editing on that wave after the fact. So literally just letting it sit there, like it was a group channel, but a pre-rendered one. Would this be ok?

[z.prime]

Why don't you just freeze tracks?

[Guenon]

I will definitely do that if rendering a subproject is a no-no. I usually run everything in realtime, not fond of Live 8's freeze.

[Guenon]

Yeah, I just tried it, I get disk overload in Live when I have just eight tracks frozen - don't really know why that is, it's always been really sub-par for me, hah. It will not buffer efficiently into RAM for some reason, and just keeps riding the disk over and over.

So, I guess I can freeze the most CPU intensive tracks if it comes to that, but I would be huuuugely happier if rendering the stuff out manually was allowed (i.e. subproject style), then chuck it onto an audio track and let it be. It buffers correctly that way, and causes no overload.

[Guenon]

I figured, as it says in the rules about bouncing tracks "Samplers used must be host samplers or freeware and easily available", if rendering stuff manually and loading it into samplers would actually be okay, then rendering stuff and placing it straight onto audio tracks is a more straightforward way of doing the same thing?

Just wanting to make sure what's okay before doing anything silly

[Guenon]

And hey! Is the OSC slack channel still alive? : D

[z.prime]

Yes, slack channel is alive periodically.

As for the multiple project thing, the only issue is "not touching audio once it leaves the daw" will have to defer to Brian on this one

[Guenon]

Yes, slack channel is alive periodically.

I really need to join that =)

As for the multiple project thing, the only issue is "not touching audio once it leaves the daw" will have to defer to Brian on this one

A clarification would be good, yeah. Loading rendered audio into a sampler is mentioned in the rules, and that seems like even more touching the audio after rendering

I'll manage with the freezing thing if it's strongly preferred, no worries. It's just, if it's fine to render something out and load it onto an audio track, I can definitely see myself doing that with Tyrell to lessen the load, later this month. (Still no promises I'm committing to a tune, but... I'm experimenting with it, and something's definitely up with this one, wink ). The way I see it, I could for example do a copy of the main project, then mute everything but drums, render the whole tune out like that, then remove all the drum instances for good, and load that wav onto an audio track, boom. Archiving the original project file with the drum Tyrells in it, of course.

[solidtrax]

Osc Slack? Where? I'm always on Slack (at work :p)!

[z.prime]

Osc Slack? Where? I'm always on Slack (at work :p)!

Check the first post in the Tyrrell thread, there's a link.