And the problem isn’t confined to instruments — high-quality, “mastering”-type audio processing plug-ins also stress CPUs to a level that if they were human, would cause them to take to the streets in protest. Even if you have the latest multi-GigaHertz pet brain from Intel, AMD, or IBM, a project that’s loaded with plug-ins is asking for trouble, particularly with low system latencies. What’s more, soft samplers often cram their samples into RAM, leaving less for system functions. If they instead stream samples from disk, then they’re placing more demands on the hard drive. Dilemma . . .
The freeze function is designed to preserve the audio that results from using these plug-ins, while dramatically reducing CPU resource consumption. It’s a clever trick, but interestingly, the ability to freeze has been available since long before companies decided to make it into a “one-click” operation. So let’s examine how the freeze function works, how to freeze with just about any host, and “à la carte” freezing for customized freeze functions.
WHY AM I FREEZING?
Simple: It’s February, and the window’s open. Seriously, though, you want to do the freeze thang because your computer’s audio is showing signs of overload: Gapping, crackling, and maybe even a complete audio engine meltdown.
The freeze process takes advantage of the fact that, thanks to high-speed/density hard drives, modern hard disk recording programs can run dozens (even hundreds) of audio tracks without breaking a sweat. So, the freeze function implements a two-step process: Convert the soft synth’s output to a hard disk audio track, then “disconnect” the soft synth from the CPU.
Different programs handle this differently. For example, some programs can’t really do anything with the frozen track, while others let you move it, or perhaps let you change signal processing being applied to the track, even though it’s “frozen.”
Those with older host versions that don’t freeze, or even newer programs that don’t have a freeze option, can nonetheless enjoy the benefits of unloading soft instruments from the CPU — and still be able to edit them later if needed. Here’s the general procedure.
1. If audio processing plug-ins follow the soft synth, decide whether to freeze these too, or insert them in the frozen track later so you can edit their settings and process the frozen audio. If the former, leave the processors alone. If the latter, bypass any instrument track processing plug-ins. Hoewever, if the signal processors consume a lot of resources, you might consider freezing the instrument/processors combination.
2. Solo the soft synth (this usually requires soloing the audio track that the instrument plays through, as well as the MIDI track driving the instrument). With a multi-timbral synth, decide whether you want to freeze each instrument separately, or a mix of all the instruments. If you want to freeze individual instruments, solo the MIDI track feeding an instrument, freeze it, then move on to the next instrument.
3. Play back the soloed track all the way through the song, and adjust its level to the maximum level short of distortion.
4. Bounce the soft synth to an audio track. If there’s no obvious bounce function, then send the synth output to an audio track and record it. Another possibility is to assign the synth out to a bus, then use that as an input to a track that can record the bus out.
5. Play back the bounced or recorded track to verify it sounds the way you want.
6. If you didn’t freeze processing along with the instrument and want to process the frozen track, copy the processor from the original track and insert it into the bounced track. If you can’t copy it, then save the processor settings as a preset, insert an instance of the processor into the bounced track, and load it with the preset parameters.
Now that our processor-leeching soft synth track is a lean, efficient hard disk audio track, we need to disconnect the instrument from the CPU. Bypassing/muting the instrument isn’t good enough, because most programs assume that you want the instrument to play back instantly if you disable bypass or mute, so they leave it “attached” to the CPU. Sonar versions prior to 4.0 (which added freeze) can use the “archive” function to disconnect the instrument, but in most other cases you’ll need to do the following.
1. Save the patch and any samples for the soft synth whose output you bounced.
2. If possible, write-protect or lock the MIDI track that drives the synth. You’ll need it again if you want to do additional editing. Note that MIDI tracks consume virtually no computer resources, so don’t worry about leaving them intact.
3. Remove the instrument plug-in from its track.
4. Document which instrument you used, which patch, and which MIDI track drives it. Your host program may have a “notepad” function for this, or you may be able to save this information in a track (e.g., as part of the name). You’ll want this info if you change your mind later and need to edit the track.
TIME TO THAW
If you want to thaw (“unfreeze”) the track and do additional editing, re-insert the soft synth into its audio track, load the appropriate patch (and samples, if applicable), and make sure the companion MIDI track is assigned to the instrument. Mute the “frozen” hard disk audio track you bounced previously (or delete it altogether if you’re feeling brave, as you’re going to be changing it anyway), make your tweaks to the soft synth’s MIDI track, and if needed, freeze again after doing your tweaks.
A REAL-WORLD EXAMPLE
Ableton Live 4.0 currently doesn’t offer a freeze function, but you can still freeze with ease. Assuming you have an instrument set up in conjunction with a MIDI track and the instrument is producing audio, start the freeze process by inserting an audio track. Set its “Audio From” field to the track playing back the instrument (e.g., if the MIDI track is named 2 MIDI, set Audio From to “2 MIDI”). Mute all other tracks, then record the instrument output into the new audio track (Fig. 1). When the track is recorded, you can remove the soft synth. The new audio track has essentially “frozen” the synth track.
A potentially simpler approach is that many host programs now offer what’s called a “one-click” freeze, because all you have to do is click on a freeze button, and all the work is handled for you automatically. For example, Fig. 2 shows a track being frozen in Sonar 4. Click on the Freeze button, and Sonar does the rest. Fig. 3 shows what happens when you freeze a track in Cubase: It’s grayed out in the VST Instruments window and the Mixer so you’re aware that it’s frozen.
Even if your host program has a freeze function, sometimes you might want to do à la carte freezing — like freezing the instrument without processors, or with selected processors. In any event, now you know enough to adapt the concept to your own needs, so I guess the appropriate signoff is — stay cool!