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 Store

Audiobus is the app that makes the rest of your setup better.

Least CPU hungry synth apps?

hey all, I'm sure this has been discussed in the past - but perhaps time to update? What in your collective opinion, are the best CPU effective iOS apps? I'm looking for best trade-off of CPU versus sound quality / versatility / power, etc.

E.g. Poison 202 seems like a very simple synth app, low CPU usage, but a huge sound. Whereas running several instances of Model 15 would probably exhaust my poor little iPad Air 2.

Cheers!

«1

Comments

  • edited February 2020

    It's a good quesion to ask ... off the top of my head from my own collection, the following seem pretty good for efficient CPU though I haven't tested them in detail:

    • Lagrange and Kronecker (IceGear)
    • Axon and Phosphor 2 (Audio Damage)
    • Ruismaker Noir and Ripplemaker (BramBos)

    Another angle is to consider how other synth apps can be run with greater CPU efficiency e.g.

    • restricting max polyphony (it's always a sign that a synth can be CPU hungry if it has limited max polyhony e.g. Continua, Factory, Model 15 etc)
    • maybe avoiding analogue modelled oscillators\effects
    • not using unison effects
    • shortening release times on pads

    Another angle is to consider efficient routing (e.g. in AUM if you use that) i.e. to reduce the number of effect app instances, use send FX busses rather than a separate insert effect instance per track etc.

  • KQ Dixie, Bleass Alpha, AudioLayer

  • Sugar-bytes Factory. Great sound, can go really out there and uses hardly any juice.

  • It all depends on what the synth is doing:

    1. how many oscillators is it running? are they running single cycle waveforms, wavetables or samples?
    2. how many filters is it running? are they digital, analog modelled? state variable, ladder, diode ladder, etc ?
    3. how many effects are you using?
    4. What's the sample rate?
    5. how many voices is running?

    So, without keeping those variables constant across multiple synths, I don't think it doesn't make much sense to compare them.

  • edited February 2020

    Sunrizer I’ve just tried 10 instances playing arpeggio patterns on hold in Audiobus at 128 frames. No issues at all. It sounded messy though, I wouldn’t want to try it again ;)

  • edited February 2020

    To my surprise, Sunrizer was more CPU-hungry than Zeeon, I wasn't even able to play 4 instances (4 voices poly each) without severe glitches in AUM @ 256 samples buffer setting. (iPad 6).
    I was even more surprised that Electribe Wave is quite an efficient AUv3 wavetable synth.

    Here's a short list of a few synths' peak CPU load, measured over one-second intervals, with the setup mentioned above:
    23% KQ Dixie
    24% Electribe Wave
    29% AD Phosphor 2
    32% bs-16i
    41% iSEM
    53% Synthmaster One
    57% Phasemaker
    67% Zeeon
    82% Nikolozi NPD

    Over 100% and crackling: Sunrizer with Reverb=off in all instances

  • I once did a little test to see how which app would allow me to open the highest number of instances without changing the default settings, and without playing any notes. KQ Dixie was the winner by far; with 44 instances loaded into AUM, my CPU was still at 65% on my 2017 10.5 inch Pro.

  • @kv331audio_bulent said:
    It all depends on what the synth is doing:

    1. how many oscillators is it running? are they running single cycle waveforms, wavetables or samples?
    2. how many filters is it running? are they digital, analog modelled? state variable, ladder, diode ladder, etc ?
    3. how many effects are you using?
    4. What's the sample rate?
    5. how many voices is running?

    So, without keeping those variables constant across multiple synths, I don't think it doesn't make much sense to compare them.

    Agree with this. We know that some synths are little CPU hungry than others, but trying to put figures on them does not work out well. I’ve come across well over 100% cpu number variations for the same synths. Too many variables. You just get to know over time which synths you can run more instances of on the particular iPad you are using.

    With an Air2, track freezing and limiting apps is just a fact of life :p

  • @Fruitbat1919 said:

    @kv331audio_bulent said:
    It all depends on what the synth is doing:

    1. how many oscillators is it running? are they running single cycle waveforms, wavetables or samples?
    2. how many filters is it running? are they digital, analog modelled? state variable, ladder, diode ladder, etc ?
    3. how many effects are you using?
    4. What's the sample rate?
    5. how many voices is running?

    So, without keeping those variables constant across multiple synths, I don't think it doesn't make much sense to compare them.

    Agree with this. We know that some synths are little CPU hungry than others, but trying to put figures on them does not work out well. I’ve come across well over 100% cpu number variations for the same synths. Too many variables. You just get to know over time which synths you can run more instances of on the particular iPad you are using.

    With an Air2, track freezing and limiting apps is just a fact of life :p

    And using built-in plugins/effects as far as possible as opposed to external ones :smile:

  • edited February 2020

    @rs2000 said:
    Here's a short list of a few synths' peak CPU load, measured over one-second intervals, with the setup mentioned above:
    23% KQ Dixie
    24% Electribe Wave
    29% AD Phosphor 2
    32% bs-16i
    41% iSEM
    53% Synthmaster One
    57% Phasemaker
    67% Zeeon
    82% Nikolozi NPD

    What's the preset settings? Is it same/similar across all plugins ? Example: 2 osc + 2 subs + 1 filter, no unison, no fx.

  • @rs2000 said:
    To my surprise, Sunrizer was more CPU-hungry than Zeeon, I wasn't even able to play 4 instances (4 voices poly each) without severe glitches in AUM @ 256 samples buffer setting. (iPad 6).
    I was even more surprised that Electribe Wave is quite an efficient AUv3 wavetable synth.

    Here's a short list of a few synths' peak CPU load, measured over one-second intervals, with the setup mentioned above:
    23% KQ Dixie
    24% Electribe Wave
    29% AD Phosphor 2
    32% bs-16i
    41% iSEM
    53% Synthmaster One
    57% Phasemaker
    67% Zeeon
    82% Nikolozi NPD

    Over 100% and crackling: Sunrizer with Reverb=off in all instances

    I’m guessing the rest of the CPU has been munched by AUM! ;)

    ❤️Audiobus

  • @Fruitbat1919 said:
    With an Air2, track freezing and limiting apps is just a fact of life :p

    not inevitably ;-) never needed freezing in Nanostudio2 on Air1, and average number of tracks in my project was always around 20-25 (mostly synth based, samoles just for drums) plus lot of FXs (reverbs, delays, eq, compressors, choruses, etc)

  • @dendy said:

    @Fruitbat1919 said:
    With an Air2, track freezing and limiting apps is just a fact of life :p

    not inevitably ;-) never needed freezing in Nanostudio2 on Air1, and average number of tracks in my project was always around 20-25 (mostly synth based, samoles just for drums) plus lot of FXs (reverbs, delays, eq, compressors, choruses, etc)

    Obviously I meant when using different AUv3 apps, as that is the subject at hand. In real life use, I can get a large track count with NS2 using its built in synth, but not with AUv3s. Multiple versions of some AUv3 apps destroy an Air2s resources quicker than I can say Model D :p

  • edited February 2020

    @kv331audio_bulent said:

    @rs2000 said:
    Here's a short list of a few synths' peak CPU load, measured over one-second intervals, with the setup mentioned above:
    23% KQ Dixie
    24% Electribe Wave
    29% AD Phosphor 2
    32% bs-16i
    41% iSEM
    53% Synthmaster One
    57% Phasemaker
    67% Zeeon
    82% Nikolozi NPD

    What's the preset settings? Is it same/similar across all plugins ? Example: 2 osc + 2 subs + 1 filter, no unison, no fx.

    That's not the way I was approaching it, I rather used the synths like a musician would, I always chose brass- and strings-like patches that somehow sounded similar.
    If there's a synth that sounds as good with 2 oscillators as another one with 4 oscillators then so be it. In case of SM One I chose four different patches from the KEY and PAD category.

    I agree that I could do a more precise shootout but I'll leave that to someone ready to spend the time.

  • @supadom said:

    @rs2000 said:
    To my surprise, Sunrizer was more CPU-hungry than Zeeon, I wasn't even able to play 4 instances (4 voices poly each) without severe glitches in AUM @ 256 samples buffer setting. (iPad 6).
    I was even more surprised that Electribe Wave is quite an efficient AUv3 wavetable synth.

    Here's a short list of a few synths' peak CPU load, measured over one-second intervals, with the setup mentioned above:
    23% KQ Dixie
    24% Electribe Wave
    29% AD Phosphor 2
    32% bs-16i
    41% iSEM
    53% Synthmaster One
    57% Phasemaker
    67% Zeeon
    82% Nikolozi NPD

    Over 100% and crackling: Sunrizer with Reverb=off in all instances

    I’m guessing the rest of the CPU has been munched by AUM! ;)

    ❤️Audiobus

    I have to say that I'm always able to use more instances in Nanostudio 2 than in AUM.
    Also battery drain is higher with AUM than with NS2.
    Glad to hear I'm not the only one 👍🏼

  • @rs2000 said:
    To my surprise, Sunrizer was more CPU-hungry than Zeeon, I wasn't even able to play 4 instances (4 voices poly each) without severe glitches in AUM @ 256 samples buffer setting. (iPad 6).
    I was even more surprised that Electribe Wave is quite an efficient AUv3 wavetable synth.

    Here's a short list of a few synths' peak CPU load, measured over one-second intervals, with the setup mentioned above:
    23% KQ Dixie
    24% Electribe Wave
    29% AD Phosphor 2
    32% bs-16i
    41% iSEM
    53% Synthmaster One
    57% Phasemaker
    67% Zeeon
    82% Nikolozi NPD

    Over 100% and crackling: Sunrizer with Reverb=off in all instances

    You need to close down all those other synth apps before firing up Sunrizer :p

  • @rs2000 said:

    I agree that I could do a more precise shootout but I'll leave that to someone ready to spend the time.

    If you read my post above, without similar preset settings, publishing numbers like that doesn't make much sense, just saying as a developer...

  • @zeroG said:

    @rs2000 said:
    To my surprise, Sunrizer was more CPU-hungry than Zeeon, I wasn't even able to play 4 instances (4 voices poly each) without severe glitches in AUM @ 256 samples buffer setting. (iPad 6).
    I was even more surprised that Electribe Wave is quite an efficient AUv3 wavetable synth.

    Here's a short list of a few synths' peak CPU load, measured over one-second intervals, with the setup mentioned above:
    23% KQ Dixie
    24% Electribe Wave
    29% AD Phosphor 2
    32% bs-16i
    41% iSEM
    53% Synthmaster One
    57% Phasemaker
    67% Zeeon
    82% Nikolozi NPD

    Over 100% and crackling: Sunrizer with Reverb=off in all instances

    You need to close down all those other synth apps before firing up Sunrizer :p

    I did. No other Synth was open before I loaded the first instance of Sunrizer.

  • edited February 2020

    @rs2000 said:

    @zeroG said:

    @rs2000 said:
    To my surprise, Sunrizer was more CPU-hungry than Zeeon, I wasn't even able to play 4 instances (4 voices poly each) without severe glitches in AUM @ 256 samples buffer setting. (iPad 6).
    I was even more surprised that Electribe Wave is quite an efficient AUv3 wavetable synth.

    Here's a short list of a few synths' peak CPU load, measured over one-second intervals, with the setup mentioned above:
    23% KQ Dixie
    24% Electribe Wave
    29% AD Phosphor 2
    32% bs-16i
    41% iSEM
    53% Synthmaster One
    57% Phasemaker
    67% Zeeon
    82% Nikolozi NPD

    Over 100% and crackling: Sunrizer with Reverb=off in all instances

    You need to close down all those other synth apps before firing up Sunrizer :p

    I did. No other Synth was open before I loaded the first instance of Sunrizer.

    I know, was kidding ;)

    @jameslondon74 if you use Cubasis then depending on what sounds you need the inbuilt Micrologue is very efficient, the same may or may not be true of Obsidian in NS2(??). Like others have said it’s bound to depend on the overall composition of the patch you’re using

  • @kv331audio_bulent said:

    @rs2000 said:

    I agree that I could do a more precise shootout but I'll leave that to someone ready to spend the time.

    If you read my post above, without similar preset settings, publishing numbers like that doesn't make much sense, just saying as a developer...

    Don't you think it's a valid approach to compare different synths' CPU load by choosing similar sounds instead of trying to match the technical architecture?
    I'm not saying that you don't make a good point but I think it's the sound that counts in the end.

  • @rs2000 said:

    @kv331audio_bulent said:

    @rs2000 said:

    I agree that I could do a more precise shootout but I'll leave that to someone ready to spend the time.

    If you read my post above, without similar preset settings, publishing numbers like that doesn't make much sense, just saying as a developer...

    Don't you think it's a valid approach to compare different synths' CPU load by choosing similar sounds instead of trying to match the technical architecture?
    I'm not saying that you don't make a good point but I think it's the sound that counts in the end.

    I can see it’s an idea, but the results become somewhat pointless when you take into account the wild variations I’ve seen. I used to do something similar and have tried many ways to get decent results, but even differing patches on the same synth can make wild variations in results.

  • edited February 2020

    I was curious too, as I’ve never paid particular attention except on cpu hogs like the Moog stuff.
    I just did a quick, unscientific test in AUM of some of my AUV3 synths .. The lower numbers are usually mono-patches or less complex polys. I tried to load the most complex patches I could find to see how high the DSP% would sustain at, using 6-7 key chords to try to max out each. (2018 ipad pro)

    Factory = 5-12%
    Bleass Alpha = 5-7%
    Mood = 5-7%
    iSem = 6-12%
    Continua 9-11%
    Phosphor = 4-12%
    digital D1 = 10-15%
    zeeon = 15-22%
    Quanta = 8-11%
    Viking = 8-12%
    Sunrizer = 3-8%
    Lagrange = 2-6%
    DRC = 8-15%
    WaveShaper = 3-8%
    EGSY01 = 2-6%
    poison-202 = 2-5%
    Synthmaster one = 5-20%
    Layr = 18-25%
    Aparillo = 6-11%
    tera = 4-24%
    kasper = 4-28%
    Volt = 5-16%
    Shockwave = 5-9%
    Enkl = 1-4%
    Syndt = 7-28%
    Tardigrain = 3-7%
    addictive Pro = 10-20%

    I was surprised at how light sunrizer runs. It briefly spiked at around 56% for a half second after loading each patch but held pretty steady at low cpu utilization afterward.
    Bleass Alpha, Tardigrain, Aparillo, Sunrizer, Poison-202, Factory, Mood, Lagrange and Continua would probably be my choices for most bang for the CPU buck.

  • edited February 2020

    @Fruitbat1919 said:

    @rs2000 said:

    @kv331audio_bulent said:

    @rs2000 said:

    I agree that I could do a more precise shootout but I'll leave that to someone ready to spend the time.

    If you read my post above, without similar preset settings, publishing numbers like that doesn't make much sense, just saying as a developer...

    Don't you think it's a valid approach to compare different synths' CPU load by choosing similar sounds instead of trying to match the technical architecture?
    I'm not saying that you don't make a good point but I think it's the sound that counts in the end.

    I can see it’s an idea, but the results become somewhat pointless when you take into account the wild variations I’ve seen. I used to do something similar and have tried many ways to get decent results, but even differing patches on the same synth can make wild variations in results.

    Yes, two sounds may sound similar but their subtle differences are heard only on high end speakers or headphones. By the time we configure presets from two different synths to match their sounds “accurately”, they should be consuming the same amount of CPU - provided they are using the same effects, etc.

  • @rs2000 said:
    To my surprise, Sunrizer was more CPU-hungry than Zeeon, I wasn't even able to play 4 instances (4 voices poly each) without severe glitches in AUM @ 256 samples buffer setting. (iPad 6).
    I was even more surprised that Electribe Wave is quite an efficient AUv3 wavetable synth.

    Here's a short list of a few synths' peak CPU load, measured over one-second intervals, with the setup mentioned above:
    23% KQ Dixie
    24% Electribe Wave
    29% AD Phosphor 2
    32% bs-16i
    41% iSEM
    53% Synthmaster One
    57% Phasemaker
    67% Zeeon
    82% Nikolozi NPD

    Over 100% and crackling: Sunrizer with Reverb=off in all instances

    The AUM percentages are only a rough guide, since the actual processor load depends a great deal on

    • patch used
    • synth fixed cost processing overheads (processing required whether playing notes or not e.g. GUI redraw, internal processing, modulation calculations etc)
    • CPU speed, since iOS boosts and throttles the CPU constantly depending on demand, so the AUM CPU meter only indicates the CPU load % at its current speed)
  • A few things:

    • you can't compare CPU efficiency, by looking at the CPU meter for an instance of a synth. The number you see may or may not be meaningful because of how throttling works. See https://wiki.audiob.us/cpu_load_comparisons_and_testing
    • you can't just straight up compare synth A to synth B if one or both of those are highly configurable in their setup. For instance, a lot of comparisons that include LayR don't take into consideration that in LayR, one can have many tens of oscillators and filters running. So, one instance of a multi-layered LayR patch might be the equivalent of running many instances of a simpler synth. A pet peeve of mine is people talking about LayR being resource heavy event though it is pretty efficient -- because people are comparing its CPU use when running a patch with a ton of layers. On my iPad Gen 6, I've had LayR running over 100 oscillators (trying to mock up the THX Deep Sound)
  • Also, synth-on-synth patch comparisons may not mean much, if one synth uses a single-cycle WAV as its oscillator, and another uses analogue modelled circuit oscillators. The former will use a fraction of the latter's processing power.

  • @MobileMusic said:

    @Fruitbat1919 said:

    @rs2000 said:

    @kv331audio_bulent said:

    @rs2000 said:

    I agree that I could do a more precise shootout but I'll leave that to someone ready to spend the time.

    If you read my post above, without similar preset settings, publishing numbers like that doesn't make much sense, just saying as a developer...

    Don't you think it's a valid approach to compare different synths' CPU load by choosing similar sounds instead of trying to match the technical architecture?
    I'm not saying that you don't make a good point but I think it's the sound that counts in the end.

    I can see it’s an idea, but the results become somewhat pointless when you take into account the wild variations I’ve seen. I used to do something similar and have tried many ways to get decent results, but even differing patches on the same synth can make wild variations in results.

    Yes, two sounds may sound similar but their subtle differences are heard only on high end speakers or headphones. By the time we configure presets from two different synths to match their sounds “accurately”, they should be consuming the same amount of CPU - provided they are using the same effects, etc.

    Sounds like a task that most likely nobody would be ready to spend such an insane amount of time. Are you?
    Also I disagree that adjusting two different synths to sound accurately similar (as far as possible, that is) will lead to similar CPU cost. The most interesting aspect IMHO is the maximum number of synths you can use in your project without crackling issues and that not only depends on the average CPU load but also very much on the CPU load variations over time. Some synths manage to distribute it fairly evenly, others eat the CPU in short peaks, leading to audio dropouts even though the overall CPU load might be at 40% average.

    Sorry that I didn't take more time for comparing, everyone's invited to do it better than me :smiley:

  • edited February 2020

    By the way, Poison-202 just got updated for iPad Pro 11” and 12.9” and iPhone 11 Pro Max resolutions with crispier graphics and 5 more presets

  • @Jocphone said:
    Sugar-bytes Factory. Great sound, can go really out there and uses hardly any juice.

    That's good to know - I have a feeling this might become my go-to synth.

  • Well, a little clarity here but not much. Seems like the question was kinda pointless. Oh well.

Sign In or Register to comment.