Loopy Pro: Create music, your way.

What is Loopy Pro?Loopy Pro is a powerful, flexible, and intuitive live looper, sampler, clip launcher and DAW for iPhone and iPad. At its core, it allows you to record and layer sounds in real-time to create complex musical arrangements. But it doesn’t stop there—Loopy Pro offers advanced tools to customize your workflow, build dynamic performance setups, and create a seamless connection between instruments, effects, and external gear.

Use it for live looping, sequencing, arranging, mixing, and much more. Whether you're a live performer, a producer, or just experimenting with sound, Loopy Pro helps you take control of your creative process.

Download on the App Store

Loopy Pro is your all-in-one musical toolkit. Try it for free today.

32 bit float-how things are today on ipados/ios

Hello everyone. I know that this topic has already been discussed. But it feels incomplete. So, I looked at the settings in Cubasis, AUM, Hokusai recorder (what I have installed). It is indicated that they support 32-bit float. But whether this is the case is a question. I plan to use it with Zoom UAC 232 (I'm thinking about buying it) in the field. But there are still doubts. I am also considering the recorder, but for now I want to understand the question about 32 bits and ipados/ios.

Comments

  • Logic recently added 32-bit float in the the 1.1 update as well. You can toggle it on in the settings.

  • nanostudio2 supports 32bit float both for export and for import into sampler …

  • edited November 2023

    well, imho you can just forget about it... it‘s marketing nonsense. >:)

    To be able to record more than 20 (valid) bits (analog to digital conversion) will set you back somewhere in the $5k range.
    (dynamic and noise figures of the amp/interface have to exceed 120dB)
    The format provides features to prevent digital overloading, but that‘s about it.
    Audio is identical to 24bit integer in best case, never „better“.

  • @bluegroove said:
    Logic recently added 32-bit float in the the 1.1 update as well. You can toggle it on in the settings.

    Thank you. They added 32-bit support to the desktop only in 10.8.

  • @dendy said:
    nanostudio2 supports 32bit float both for export and for import into sampler …

    Thank you. I wonder if anyone has checked the 32-bit floating entry.

  • @Telefunky said:
    well, imho you can just forget about it... it‘s marketing nonsense. >:)

    To be able to record more than 20 (valid) bits (analog to digital conversion) will set you back somewhere in the $5k range.
    (dynamic and noise figures of the amp/interface have to exceed 120dB)
    The format provides features to prevent digital overloading, but that‘s about it.
    Audio is identical to 24bit integer in best case, never „better“.

    Would you agree that managing the overload of the recording level is also not bad? Just an interesting topic (I can't give an accurate assessment of this yet until I try it myself). I've heard about problems with the headphone amplifier on the same Zoom that were eventually solved.

  • Well, if you consider your mix, you have to do the gain staging anyway.
    (which is pretty easy today because regular 24bit provides more than enough headroom and there are tools to assist you, that didn‘t exist decades ago)

  • @Telefunky said:
    well, imho you can just forget about it... it‘s marketing nonsense. >:)

    To be able to record more than 20 (valid) bits (analog to digital conversion) will set you back somewhere in the $5k range.
    (dynamic and noise figures of the amp/interface have to exceed 120dB)
    The format provides features to prevent digital overloading, but that‘s about it.
    Audio is identical to 24bit integer in best case, never „better“.

    One big advantage I see in floating point recording is the fact that you don't have to worry about clipping at all at the software level, so you can probably run the interface much "hotter" and not fiddle with Gain controls all the time. Granted, there'll still be clipping at the hardware level if you go completely crazy, but that's just a fact of physics 😁

  • @solncekreeze Audio Evolution Mobile Studio added support for this a few updates back.

  • This is a hot topic now that the wireless mics (RØDE, DJI) and cameras are adding 32 bit float audio recording.

  • @SevenSystems said:
    One big advantage I see in floating point recording is the fact that you don't have to worry about clipping at all at the software level, so you can probably run the interface much "hotter" and not fiddle with Gain controls all the time.

    Why would one run an interface „hotter“, when it‘s performing totally linear ? o:)

  • @Telefunky said:

    @SevenSystems said:
    One big advantage I see in floating point recording is the fact that you don't have to worry about clipping at all at the software level, so you can probably run the interface much "hotter" and not fiddle with Gain controls all the time.

    Why would one run an interface „hotter“, when it‘s performing totally linear ? o:)

    Because the A/D converters have a noise floor.

  • edited November 2023
    The user and all related content has been deleted.
  • edited November 2023

    Because the A/D converters have a noise floor.

    Of course they have... but not that relevant, compared to the noise floor of the input stage.

    Which brings us back to square one:
    my Audient ID22 has a -90dB noise floor under best conditions on it‘s mic input.
    Which is less than fits in a 16bit data word. Line input may perform a little better, but budget outboard gear usually carries even more noise.
    There is nothing that isn‘t covered by 24bit integer regarding convenient headroom.
    Or rather it won‘t get any „better“ in the float format because the noise is simply too high.

    On synthetic sounds this doesn‘t apply to the same degree, but your ears will burst at levels when it might actually matter.

  • @Telefunky said:

    Because the A/D converters have a noise floor.

    Of course they have... but not that relevant, compared to the noise floor of the input stage.

    Good counter! I'll accept that 😄

    I think we can agree on the fact that neither the dynamic range nor the noise really matter here -- but the big advantage of not having to care about setting the right input level, because there's no point at which the digitized signal can clip, is still relevant. Basically, this unifies the signal format of the "input stage" of your workflow with the format that the DAW works with internally anyway. So your whole workflow is clip- and worry-free.

  • edited November 2023

    @SevenSystems said:
    ... Basically, this unifies the signal format of the "input stage" of your workflow with the format that the DAW works with internally anyway. So your whole workflow is clip- and worry-free.

    I‘m not sure about DAW formats internally... Pro Tools had 48/56 (?) bits internally decades ago, some modern stuff claims 64bit, etc.
    (but I never cared enough to bother with the details) o:)

    I just wanted to provide some real world figures to the buzzword.
    It will sell like hot cake anyway :mrgreen:

  • @dendy said:
    nanostudio2 supports 32bit float both for export and for import into sampler …

    But, I don't think it supports 32bit float recording, which I think is the original topic. I don't think it was an option when NS2 was last updated.

    Sorry to be nit-picky. I just didn't want anyone to get the wrong idea that NS2 can record at 32 bit from an external interface.

  • edited November 2023

    @wim said:

    @dendy said:
    nanostudio2 supports 32bit float both for export and for import into sampler …

    But, I don't think it supports 32bit float recording, which I think is the original topic. I don't think it was an option when NS2 was last updated.

    Sorry to be nit-picky. I just didn't want anyone to get the wrong idea that NS2 can record at 32 bit from an external interface.

    32 bit float recording must be supported by HW interface in first place, otherwise it's just fake - upsampled recorded material by application ..

    In case you use sound card which supports 32bit float input then NS2 should record in 32bit float input .. At least this is the way how it is implemented from NS point of view. I am not sure how iOS handles external sound devices - but this will be same for all iOS apps.. NS internally processes ALL audio in 32bit floating point format, so even if you load there lower bitrate audio, it upsamples it for internal processing to 32bit f.

    If iOS really limits recording stream let's say to 24bit fixed point then well, none of iOS apps would be able to utilised high resolution HW directly. No idea what is situation, maybe some dev like @brambos or @j_liljedahl may put more light on this .

    To put it to another context- if some app records 32bit float from your iPad microphone, it just means it recorded 16bit/48k audio and then allows you to save it resampled it to 32bit float / 48 khz .. which is something what doesn't make much sense, but yes, It can be done also in NS (you record audio and then you export it as 32bit float)

  • wimwim
    edited November 2023

    Sorry @dendy but I think you have incomplete / incorrect information on this matter. Support for 32 bit internal processing has of course always been there in iOS and in NS2. However support for recording at 32 bits from 32 bit capable external interfaces has not been possible on iOS until recently and has to be specifically implemented in each host. It has only recently begun to show up in DAWs. I believe Cubasis was the first DAW to do this some time earlier this year. Audio Evolution Mobile is another DAW that recently added this support.

    I don’t have the time to back-trace my sources for this info right now, but there is a thread where this was discussed earlier should you care to verify. Or continue to assume I’m incorrect. Doesn’t matter to me.

    It isn’t very important as regards NS2 as without audio tracks I doubt those who are concerned with this topic are using NS2 much for heavy audio tracking anyway.

  • @dendy said:

    @wim said:

    @dendy said:
    nanostudio2 supports 32bit float both for export and for import into sampler …

    But, I don't think it supports 32bit float recording, which I think is the original topic. I don't think it was an option when NS2 was last updated.

    Sorry to be nit-picky. I just didn't want anyone to get the wrong idea that NS2 can record at 32 bit from an external interface.

    32 bit float recording must be supported by HW interface in first place, otherwise it's just fake - upsampled recorded material by application ..

    In case you use sound card which supports 32bit float input then NS2 should record in 32bit float input .. At least this is the way how it is implemented from NS point of view. I am not sure how iOS handles external sound devices - but this will be same for all iOS apps.. NS internally processes ALL audio in 32bit floating point format, so even if you load there lower bitrate audio, it upsamples it for internal processing to 32bit f.

    If iOS really limits recording stream let's say to 24bit fixed point then well, none of iOS apps would be able to utilised high resolution HW directly. No idea what is situation, maybe some dev like @brambos or @j_liljedahl may put more light on this .

    No, I don't have a lot to add to the conversation other than that all my apps use CoreAudio's native 32 bit format. Which means that everything that comes in and goes out of the apps is in 32 bit - and stays 32 bit inside my apps too.

    I can imagine that recording and running your audio hardware in 32 bit as well, at the very least avoids having to convert formats twice, which maybe helps getting slightly lower latency?

    I have nothing to confirm this however, just speculating.

  • The use and value of 32-bit float sample data varies depending on where it is in the chain.

    I’d be very surprised to find any software using a fixed point format for internal processing in 2023! More than anything this is due to use of CPUs which handle this well, algorithms can be a little simpler (less watching headroom! And the scaling needed to manage that) and so likely faster.

    Storage likely doesn’t matter much, but does allow absolute level to be ignored - though as absolute level is relevant for non-linear processing that reduces the sense in this!

    For recording 32-bit float is used by hardware to avoid clipping using a “trick” - have essentially 2 converters in parallel, the main one optimised for normal levels so 0dBFS where you’d expect and I assume you’d still set level as normal. But… if the signal exceeds 0dBFS / clips the main converter the secondary converter run at lower gain is switched in. This will be great in situations where occasional “too loud” signals occur but dynamic range says the input gain shouldn’t be lowered.

  • edited November 2023

    @MadGav said:

    For recording 32-bit float is used by hardware to avoid clipping using a “trick” - have essentially 2 converters in parallel, the main one optimised for normal levels so 0dBFS where you’d expect and I assume you’d still set level as normal. But… if the signal exceeds 0dBFS / clips the main converter the secondary converter run at lower gain is switched in.

    The Rode NT1 5th Gen uses 4 internal pre-amps/ADC's. Dont' know how they are 'gain staged' but if they use 4 of their revolution preamps they could theoretically have enough juice to amplify the noise-floor of the mic to 0dBFS and at the same time pad it down enough so when a 140db SPL signal hits the capsule it would still have enough headroom to not clip which should cover most normal and abnormal situations.

    The noise-floor in my room is around ~40dB(a) SPL so an NT1 would not improve the noise situation at all compared to my current combo using the Audient ID4 Mk2 and Lewitt LCT441Flex which has a self noise of 7dB(a) and most of the noise is caused by the actual room and not the equipment :sunglasses:

    The just about the only advantage of the NT1 5th Gen for me would be that I would not have to worry about setting the gain and could only focus on placing the mic in the desired location and hit record.
    (With that said I am considering getting one just to get comfortable with using 32-bit float recording technology).

    In the end I see 32-bit float recording like HDR photography.

    Cheers!

  • edited November 2023

    @brambos said:
    No, I don't have a lot to add to the conversation other than that all my apps use CoreAudio's native 32 bit format. Which means that everything that comes in and goes out of the apps is in 32 bit - and stays 32 bit inside my apps too.

    and it uses 32bit floating point or fixed point (int) format ?

    Sorry for me being lazy ass to google it o:)

  • @dendy said:

    @brambos said:
    No, I don't have a lot to add to the conversation other than that all my apps use CoreAudio's native 32 bit format. Which means that everything that comes in and goes out of the apps is in 32 bit - and stays 32 bit inside my apps too.

    and it uses 32bit floating point or fixed point (int) format ?

    Sorry for me being lazy ass to google it :-)

    CoreAudio use 32-bit floats internally if I'm not mistaken :sunglasses:

  • edited November 2023

    @Samu said:

    @dendy said:

    @brambos said:
    No, I don't have a lot to add to the conversation other than that all my apps use CoreAudio's native 32 bit format. Which means that everything that comes in and goes out of the apps is in 32 bit - and stays 32 bit inside my apps too.

    and it uses 32bit floating point or fixed point (int) format ?

    Sorry for me being lazy ass to google it :-)

    CoreAudio use 32-bit floats internally if I'm not mistaken :sunglasses:

    Expecting this is the case ...

  • Float yes.

    I haven't touched fixed point in a decade (and it was no fun back then so it wouldn't be in 2023)

  • It still is a lot of fun with the Lexicon Vortex (ADSP2105, 16bit fixed point DSP) o:)

  • @Telefunky said:
    It still is a lot of fun with the Lexicon Vortex (ADSP2105, 16bit fixed point DSP) o:)

    The quantization noise gives the old gear some real 'character' :sunglasses:

  • AUMs 32 bit float recording option is only relevant for the recorded file format. Internally it always uses 32 bit floats. And in case you have an audio interface that supports it, you would get 32 bit float hardware in/out as well. But my point is that even with a 16 bit audio interface, recording in 32 bit float might make sense if you're doing any processing and/or using any plugin instruments etc, mostly because of the ability to record signals above 0dBFS without clipping. Resolution wise, 32 bit float has similar resolution to 24 bit int (because float format is divided in mantissa and exponent), so it's rather the dynamic range that is the win here.

  • edited February 22

    Any experts here wanna weigh in on the iPhone app “Stereo Microphone” (by Ryouta Kira) and its 32-bit float capability? I did a brief test recording acoustic cello pizzicato and purposefully overloading my iPhone 12 pro’s front mics, then opened the file on Audio Evolution Mobile Studio (also set to 32-bit float) to normalize the audio at a lower volume (-1db). The clipping disappeared, audio levels no longer going into the red. https://apps.apple.com/us/app/stereo-microphone/id6449208745

Sign In or Register to comment.