Audiobus: Use your music apps together.
What is Audiobus? — Audiobus is an award-winning music app for iPhone and iPad which lets you use your other music apps together. Chain effects on your favourite synth, run the output of apps or Audio Units into an app like GarageBand or Loopy, or select a different audio interface output for each app. Route MIDI between apps — drive a synth from a MIDI sequencer, or add an arpeggiator to your MIDI keyboard — or sync with your external MIDI gear. And control your entire setup from a MIDI controller.
Download on the App StoreAudiobus is the app that makes the rest of your setup better.
Tips on keeping low DSP?
![[Deleted User]](https://secure.gravatar.com/avatar/a3dfdd5b0ba6b0f65415443199997ae4/?default=https%3A%2F%2Fvanillicon.com%2F5d1df3dd5891755bc0d00de3da842543_200.png&rating=g&size=200)
Are there any tips on keeping low dsp levels?
Does adding peripherals (audio interface, midi controllers etc) have an effect?
Running new ipad pro 2018, latest ios, with simple au setups in apematrix and aum.
-thx
Comments
Samplerate and Buffersize will have by far the largest impact on DSP performance.
Lower Samplerate = less work = better performance but also higher latency
Higher Samplerate = more work = worse performance but also lower latency
with the buffer size it's the opposite.
Higher buffer size = more efficient = better performance but also higher latency
Lower buffer size = less efficient = worse performance but also lower latency
Hope that helps.
Just use acoustic instruments, or if you must be electronic, go for analog stuff.
😉
Really @ka010 has mentioned the most important stuff.
My question to the knowledgeable...
Does using an USB Audio Device take some processing load off of the main Processor?
Yeah thx for that, been making those adjustments. Was wondering of any further. Also a bit suprised any adjustments need to be made considering these minimal setups.
-Cheers
@hibjshop : apps and plug-ins vary widely in how much DSP they use. Some analog synth emulations (such as the Moog apps) are DSP heavy (but sound great). Reverbs and room simulations tend to be DSP heavy. So using them judiciously is wise.
Sample rate doesn't normally affect the latency - it affects the sound quality. 99% of the time there's really no need to touch this, unless it relates to a plugin doing oversampling. 44/48K will be fine.
Buffer size is the one to play with. Also you can put the ipad/iphone into airplane mode, and make sure you've closed all other apps.
Adding peripherals shouldn’t have noticeable effect. The quality of those peripherals will affect how they perform, but shouldn’t affect CPU usage. As mentioned above, buffer settings and sample rate are the main things you can play with - primarily buffer settings. If you’re not playing live then set ‘em high. If you’re playing live then lower buffer settings will reduce latency. Set ‘em as low as you can without getting crackling.
I believe the first app you load is going to determine the buffer settings. So, best to start with the host, like AudioBus, AUM, your DAW, etc. I could be wrong on this point though.
Sample rate settings are less direct and more unpredictable to vary. I would focus on buffer settings over sample rate. Ymmv.
Above all, don’t place too much stock in any CPU meter readings you see. That isn’t an exact science on iOS, and different apps report differently. The only real measurement is whether you’re getting crackles or dropouts or not. If you’re not getting them, and the latency is acceptable, my advice is to forget about the meters. Seriously. Forget about them. Their only use is in helping to narrow down which plugins are causing issues if you have them.
great info, much thx!
Sure it does, higher sample rate means a shorter interval in between render cycles and that equates to less latency.
For instance:
44.1khz and 256 samples you get 256/44100 = 5.8ms
bump it up to 96khz and 256 samples and you get down to 2.6ms
Sure, but it would be an eccentric way to try and reduce latency. And in practice you're going to end up increasing your buffer size to at least 512.
Either way if you're having problems with crackling, change the sample rate to 48/44.1 and then don't touch it. Could try switching between those I guess - very occasionally that can fix things.
I'm sorry but that makes no sense.
I have an interesting (but horribly filmed) video that I believe to be relevant. You can see that just by opening RE-1's UI inside AUM, the DSP load gets greately reduced, from close to 100% to around 20%. I didn't think of recording the audio (it was a constantly running arpeggio), but when the load is near 100% there's constant crackling, which disappear once the reported DSP load gets down to 20%.
I've read about the existence of high and low (priority? load?) cores in the new iPad Pros, and a theory that in certain cases the audio load could be "pushed" to the high cores. This seems like an example of that behavior.
This is a 12.9" 2018 iPad Pro with iOS 12.1.4. It happens with or without an external audio interface, and with “Limit Frame Rate” on and off.
I'm seeing the same behavior with any AU with rather “active” UI, like RE-1 and Rozeta Collider (or even KB-1, but only while pressing keys).
https://www.youtube.com/watch?v=spJe6AUZbJ0
All these CPU/DSP figures aren't about actual processing tasks (what the thing is doing), but about claims on time slices (when a task is supposed to act).
A high meter doesn't indicate a demanding calculation. Instead it tells something is begging attention.
Got it, but they’re still related, aren’t they? More high demand calculations wpuld fight for available CPU resources and make them more likely to be “late”, right?
Regardless, it’s still weird that, in my case, that figure is reduced by adding what is supposed to be more load (UI rendering).
It seems weird at first, but keep in mind that all CPU cores can and will be throttled (i.e. run at less than their max speed). So it could be that opening the GUI tickles a throttled core and forces it to start working at full speed, causing the overall percentage of CPU load to go down (because more power suddenly becomes available).
Tips.
I've done testing on quite a few plugins. Here are some that have low CPU and no spikes:
-Many of the Icegear's especially when you aren't using too many effects and in mono modes.
-Ruismakers
-For a really complex synth, Virsyn Tera is often pretty efficient. Again, esp in mono and with fewer effects.
Got it, this makes perfect sense. I’m guessing that this doesn’t happen in my regular iPad because it’s harder to push the Pro cores out of throttling. Wish there was some setting that would prevent throttling.
Yeah, that's what I'd attribute this to as well, seeing this kind of behaviour regularly during testing, most notably on the latest gen pros. Never had it cause any glitching but it does seem they're throttling very aggressively.
Agreed, Icegear and Brambos two of my favs 👍🏼
So I'm hitting too high DSP levels in AUM because my song has evolved to many tracks.
Since freezing is no option in AUM and I'm also not able to change buffer size or sample rate (IPAD Pro 2020), what other options are there to make some extra space?
How to temporarily "disable" a track? Can I disable a synth on top of the AUM track, by swiping it to the left?
@Bob Try to use instruments and FX with low CPU load. AUM isn't a classic DAW with features like track freeze so workarounds like recording channels separately and loading the audio into AUM loop players do work well but they're no fun IMHO.
I know nobody wants them anymore but old IAA apps designed for older iPads are mostly low cpu load. Samplr, Sunrizer, iSEM etc...some of them have become AUs but kept their low cpu heritage like Sunrizer, iSEM and Turnado. The old Magellan can be used as a multi effect unit and to me it sounds lush for such a light footprint app, all midi controlled to boot.
Audio Damage Replicant 2 seems very low cpu as well.
I guess it’s all trial and error in your particular configuration.
Generally though, as pointed out before, effects can be heavier than synths.
I’m surprised that nobody talks about ram. Last time I’ve checked I was at 95% while the CPU was running at a leisurely 30%.
You might not be able to freeze tracks, but you can sample your cpu-hungry synths. If there's a bass that you love in Model D, or something like that.
Thanks everyone! Well the damage is done, not able to go back in my song arrangement, unless I recover a previous version or start over. NAH! I will do some editing here and there, start recording some tracks and use the AUM file player.
These mistakes sure help in getting things done on a more efficient way the next time.
@supadom Neither free ram amount nor average cpu load will tell you how efficient a plugin is. The only reliable way to find out is to load many of them until you get dropouts.
You can record the outputs on AUM and reload the waves... It will act like freezing
Since arrival of Audiobus on iOS I’ve been running set ups at 100% minus 15% capacity of the system. The apps mentioned above are super light weight from experience
Why doesn’t cpu rise relative to DSP?
It does but you can't simply extrapolate.
👆👆👆👆 +1
Peep some older apps. I still wonder why some apps melt my cpu, when using an old AUFX or Holderness app works fine. 🤷🏼♂️
Drambo is extremely CPU efficient even though it's a rather new app. It's not about age but rather about coding quality in terms of performance.
It even runs fairly well on my old iPad Mini 2.