How to record plug-in synth audio in real-time?

Hi everyone,

The other night I was playing with some soft-synths and really getting into the groove with an evolving pad. I wanted to record the audio output as I play, to capture the exact performance, rather than record the midi and bounce to audio. My concern is the peeks and swells and general evolvy-ness of the sound will not be the same when played back, as the synth parameters are reset when restarted. At least, that's what seems to be happenning to me.

Am I missing something?

Wow, I'm surprised that a piece of software this sophisticated would have such a huge limitation. I just got on to post a question about this very topic and it was almost the top one on the forum. So, just so I hear you right, there is no way to route a midi track (after its been recorded) to an audio track? I'd like to put some effect on it like a real audio track....

thanks for any advice

Sabbathack: after you record a midi track, you can route it to any softsynth you want. You can put effects into the synths FX bin, or patch the synth track to a bus, etc. etc. You can also freeze or bounce a softsynth and convert it into an audio track.

What you CAN'T do in SONAR is play a softsynth on your midi input device (keyboard) and simultaneously record the resulting AUDIO (as opposed to MIDI) data.

Millsyy2k: a bouncedown from a MIDI track MIGHT sound exactly the same as your performance. It depends on the softsynth. Many synths include modeling or randomization parameters, or can run an oscillator from their internal clock rather than SONAR's. In those cases, the MIDI data alone isn't sufficient to recreate exactly a specific performance.

Example #1: Say you create a synth patch in which the LFO modulates a filter. You clock the LFO via the softsynth's internal clock rather than SONAR's (e.g., because you don't want the LFO directly tracking your piece's beat). As you play, the LFO-modulated filter produces a nice emphasis on a specific harmonic of a specific note at a specific moment, whch you then build on in your improvisation. Problem: since the LFO is being clocked internally by the soft-synth, meaning it is independent of SONAR's clock, the waveform will not necessarily align with your notes the same way when you re-play the file, meaning the filter will affect the notes differently.

Example #2: You use a sampler like GPO that allows for randomization of pitch on stringed instruments, to simulate the slight variations inevitable in a human performance of an instrument. Again, the specific randomization pattern isn't captured as MIDI data, so it will always be somewhat different each time you replay the file. If you're using enough randomization for it to be audible (e.g., as an effect), again you can't capture a specific performance you might particularly like. This is a particularly annoying limitation when using synths that are pure randomized generators, e.g. some Reaktor instruments, where the MIDI data might consist of nothing but "note on" followed eventually by "note off."

What bugs me is that this would seem to be a TRIVIALLY easy thing for CW to fix. All that's needed is the ability to have a synth track serve as the INPUT for an audio track. Arm the audio to record, play the synth, and away you go.

aaronk you nailed on the head with a perfect, articulate description - thank you!

I looked when I upgraded to 7 producer to be able to use a soft-synth out as an audio in. That's what I figured to be the perfect solution. I'm REALLY bummed it's not there, as MIDI improvements were one of the big selling points for version 7.

Now, I still can't capture a really intersting ambient improv on a soft synth properly. :(

ORIGINAL: smallstonefan

Hi everyone,

The other night I was playing with some soft-synths and really getting into the groove with an evolving pad. I wanted to record the audio output as I play, to capture the exact performance, rather than record the midi and bounce to audio. My concern is the peeks and swells and general evolvy-ness of the sound will not be the same when played back, as the synth parameters are reset when restarted. At least, that's what seems to be happenning to me.

Am I missing something?

Maybe, if you're filtering midi events, or quantizing in real-time.

What you're hearing, in real-time, when you play your softsynth, is MIDI going from your controller into your soft synth, and played through whatever audio engine you have installed on your pc. It's all MIDI first. Whether you record the midi as you're playing and then run in through the synth (Bounce), or capture the digital audio from this process, you're going to hear exactly the same thing, that is, unless you're filtering, or quantizing in real-time.

But, if you really want to record the audio output of your softsynth while playing "live", then send the softsynth audio out to your board and bring it back through an audio in. If you can stay in the digital realm, then you'll have the exact same result (minus the latency of your board) that you get from Bouncing--you can fixup the delay in an audio editor, if need be. But, if you have to go out via analog Audio and return to Sonar via an A/D conversion, then you will notice a difference in quality, depending on how good your converters really are. Perhaps this is a good test?

Now, I still can't capture a really intersting ambient improv on a soft synth properly. :(

why not? just press the arm automation button and all your movements will be recorded.

Just plug a cable from you soundcards output to the input and be careful to turn realtime monitoring OFF!

ORIGINAL: aaronk

Sabbathack: after you record a midi track, you can route it to any softsynth you want. You can put effects into the synths FX bin, or patch the synth track to a bus, etc. etc. You can also freeze or bounce a softsynth and convert it into an audio track.

Ahhhhh thank you, it would seem my issue is different from the original post. My issue is that I'm DuMb and don't know what I'm doing yet. ;) Thanks for the info, it was a well thought out and written response.

ORIGINAL: aaronk

Just plug a cable from you soundcards output to the input and be careful to turn realtime monitoring OFF!

Cool. Now I can't hear what I'm doing.

Well, that depends on your soundcard and setup. Most audio interfaces include a headphones output, or a simple mixer that can be used to route the softsynth track to more than one place. If your soundcard has more than one pair of outputs, you can use a Sonar bus to route the track to your soundcard's input AND your monitors.

quote:


Well, that depends on your soundcard and setup. Most audio interfaces include a headphones output, or a simple mixer that can be used to route the softsynth track to more than one place.


right, thats also why we have aux sends in sonar.

But instead of a work-around, why doesn't CW simply allow an Audio track to record the output of a Synth track? It's so bloody OBVIOUS . . .

I appreciate the input from everyone. Some still don't fully understand the problem though. aaronk really did explain it best. I have a Delta 1010. I will explore routing through the sound card digitally and see if I can do that with the 1010. Beyond that, I agree with aaronk in that this should be a feature in the product instead of requiring work arounds that may not even be applicable to some sound cards.

ORIGINAL: smallstonefan
I have a Delta 1010. I will explore routing through the sound card digitally and see if I can do that with the 1010.

I'm pretty sure the 1010 can do it. I have the Delta 2496 and it can.

Open the delta control panel and go to the monitor mixer tab. Bring up the faders on whatever channels you have your softsynth routed to--this will also send the softsynth to the monitor mixer output.

Now, in Sonar set up a new audio track and set its input to the Delta Mon. Mixer. You should be able to record the synth output now.

But yeah, Sonar should just be able to do it on its own.

It doesnt
Cj

Depends on the synth you're using. If it uses acoustic modeling algorithms, storing the midi data alone won't be enough to reproduce the same sounds. This is because those algorithms include a fair amount of uncertainty (e.g. "if x, do either y or z"), in order to model real-world physical events (which are inherently somewhat indeterminate).

The midi data is like musical notation. Writing down a middle C for violin will get you more or less the same sound from any violinist. But it won't be as exact as recording the violin directly.

ORIGINAL: ba_midi

Options->Project-MIDI

There's an option to "reset controllers" etc.... uncheck that.

What does this do?

+3 for the 'tape it' vst. insert it in the fx bin after your softsynth & hit record, then start playing back your track. drag the recorded wav back into your project afterwards. much easier than mucking about with the soundcard & track routings.