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.

Cubasis: Rendered Audio is always a bit too long

Whatever I try, the length of rendered audio from MIDI patterns in Cubasis, always seems to be a bit too long.

For example, 8 bars of MIDI create 16.011 seconds of Audio.

This seems always be the case.

And I am not sure if this is just a small piece of audio before the recording, after the recording or if the recording was stretched a bit.

Anyway, this is not good!

Any trick to prevent this?

Or is this a bug, @LFS

«13

Comments

  • I'm guessing it's the latency setting, being added as a silence at the start of the file. Auria Pro used to do this too, until the bug was fixed about a year ago.

  • I've noticed this varies from 'app to app'. Some synths create longer tails than others.
    The wave's may look 'quiet' but when normalising the seeming 'silent' portion turns out it's not so silent after all...

    Could it be that Cubasis 'waits' until the synth no longer produces any output and then stops?

  • If you do not want excessive silence at the end which can occur from long and almost silent FX tails, before exporting via mixdown set the loop markers at the start and end of your track/loop, then in the mixdown screen select the 'Mixdown Between Locators' option and turn off 'Include effect tail'.

  • I use freeze most of the time as it's faster and doesn't interrupt the flow in the way that the 'mix down' things does.
    After freezing I trim the frozen audio.

  • I don't like fade in-outs (matter if taste) so I extend the tracklength beyond the final point where everything has decayed naturally, sometimes by a dummy silent region to force the DAW to include the extension silence.
    But the 'wrong' rendering length might turn into a problem if the target is a fixed beat length of sections/stems to be used in other arrangements.

  • @richardyot said:
    I'm guessing it's the latency setting, being added as a silence at the start of the file. Auria Pro used to do this too, until the bug was fixed about a year ago.

    In a short test, I got 7.999 seconds instead of 8 in Auria Pro, as seen by AudioShare.
    But that was a different MIDI file, so i will test this again.

  • @Samu said:
    I've noticed this varies from 'app to app'. Some synths create longer tails than others.
    The wave's may look 'quiet' but when normalising the seeming 'silent' portion turns out it's not so silent after all...

    Could it be that Cubasis 'waits' until the synth no longer produces any output and then stops?

    I did select not to include the trail at mixdown

  • @AndyPlankton said:
    If you do not want excessive silence at the end which can occur from long and almost silent FX tails, before exporting via mixdown set the loop markers at the start and end of your track/loop, then in the mixdown screen select the 'Mixdown Between Locators' option and turn off 'Include effect tail'.

    Yes, i used exactly that

  • tjatja
    edited July 2018

    @Samu said:
    I use freeze most of the time as it's faster and doesn't interrupt the flow in the way that the 'mix down' things does.
    After freezing I trim the frozen audio.

    I did not yet compare freeze and mixdown, will do so.

    But anyway, to trim correctly, one would need to know if there are frames added at the front or at the tail.

    And as i wrote, it could also be, that the audio is slightly stretched!

  • I also tried BeatMaker 3, and strangely, this App does not seem to be able to calculate the size of a track by itself - you need to enter the number of bars.

    The positive effect is, that this produced the correct output length!
    So, exactly 8.000 or 16.000 seconds

  • edited July 2018

    @tja you remind me of one of our clients, each time we updated their website, they actually took a wooden ruler and measured the lengths of various layout elements on the site and called us in a panic attack, noting that the header was now 0.1mm too short!!!! ;) :D

    No, but obviously you're right and it's important if you want perfect loops exported etc...

  • edited July 2018

    @tja said:

    @AndyPlankton said:
    If you do not want excessive silence at the end which can occur from long and almost silent FX tails, before exporting via mixdown set the loop markers at the start and end of your track/loop, then in the mixdown screen select the 'Mixdown Between Locators' option and turn off 'Include effect tail'.

    Yes, i used exactly that

    If your BPM is not 60, 120 or 240 then you will always have issues...the reason is because of math...

    Sampling frequencies are in seconds, so 44.1 Khz means 44,100 samples per second...
    BPM is also in seconds...... @ 120 BPM you 2 beats per second so each beat will be 22,050 (44,100 / 2) samples long....

    If your BPM is 130 then each beat is 0.4615386 of a second, now the math isn't so simple, each beat is now 20,353.908 samples ....and you cannot have .908 of a sample so rounding is needed, it has to either become 20,353 or 20,354 samples which causes the differences in loop length.

    Can I go back to not thinking too much now :D

  • edited July 2018

    once you start dealing with this scale of 'precision' (display) be prepared for a lot of surprises regading rendering files from different apps in different formats ;)
    (or vice versa on import)
    And you don't even know if a successor app (further on in the process) will calculate on the number displayed or on it's own measuring of sample content.

  • @Telefunky said:
    once you start dealing with this scale of 'precision' (display) be prepared for a lot of surprises regading rendering files from different apps in different formats ;)
    (or vice versa on import)
    And you don't even know if a successor app (further on in the process) will calculate on the number displayed or on it's own measuring of sample content.

    120 bpm is the default in most apps for a reason ;)

  • @AndyPlankton said:

    @tja said:

    @AndyPlankton said:
    If you do not want excessive silence at the end which can occur from long and almost silent FX tails, before exporting via mixdown set the loop markers at the start and end of your track/loop, then in the mixdown screen select the 'Mixdown Between Locators' option and turn off 'Include effect tail'.

    Yes, i used exactly that

    If your BPM is not 60, 120 or 240 then you will always have issues...the reason is because of math...

    Sampling frequencies are in seconds, so 44.1 Khz means 44,100 samples per second...
    BPM is also in seconds...... @ 120 BPM you 2 beats per second so each beat will be 22,050 (44,100 / 2) samples long....

    If your BPM is 130 then each beat is 0.4615386 of a second, now the math isn't so simple, each beat is now 20,353.908 samples ....and you cannot have .908 of a sample so rounding is needed, it has to either become 20,353 or 20,354 samples which causes the differences in loop length.

    Can I go back to not thinking too much now :D

    I tested with 120 BPM.
    So, that should be an even number, right?

  • @SevenSystems said:
    @tja you remind me of one of our clients, each time we updated their website, they actually took a wooden ruler and measured the lengths of various layout elements on the site and called us in a panic attack, noting that the header was now 0.1mm too short!!!! ;) :D

    No, but obviously you're right and it's important if you want perfect loops exported etc...

    This occurred to me, as i tried to prepare a comparison of DAWs and their rendering differences.

    But i could not even get samples of the same length from them, which disables any comparison.

    And as i said, if we knew for sure if that is an empty addition at the start or at the end, we can just drop.

    But this could also be a stretching, which would make things more complicated.

  • @tja said:

    @AndyPlankton said:

    @tja said:

    @AndyPlankton said:
    If you do not want excessive silence at the end which can occur from long and almost silent FX tails, before exporting via mixdown set the loop markers at the start and end of your track/loop, then in the mixdown screen select the 'Mixdown Between Locators' option and turn off 'Include effect tail'.

    Yes, i used exactly that

    If your BPM is not 60, 120 or 240 then you will always have issues...the reason is because of math...

    Sampling frequencies are in seconds, so 44.1 Khz means 44,100 samples per second...
    BPM is also in seconds...... @ 120 BPM you 2 beats per second so each beat will be 22,050 (44,100 / 2) samples long....

    If your BPM is 130 then each beat is 0.4615386 of a second, now the math isn't so simple, each beat is now 20,353.908 samples ....and you cannot have .908 of a sample so rounding is needed, it has to either become 20,353 or 20,354 samples which causes the differences in loop length.

    Can I go back to not thinking too much now :D

    I tested with 120 BPM.
    So, that should be an even number, right?

    Should be yes.

  • I created a new, simple MIDI at 120 BPM and 1 bar, the. froze it (that included the trail) and then made a mixdown.

    Both resulting files are slightly over 2 seconds:

  • edited July 2018

    @tja said:
    This occurred to me, as i tried to prepare a comparison of DAWs and their rendering differences.

    But i could not even get samples of the same length from them, which disables any comparison.

    the infamous ever recurring null-test scenario o:)
    Imho just bail out - it's not worth the effort, as not a single person would use a DAW for just adding stems. It's all about busses and processing (fx) chains, which you cannot compare due to their different heritage.

    I recently experienced how failures during internal data transfer can spoil a signal: a well respected plugin suddenly introduced a lot of digital mud - and occasional glitches indicated that something must be awefully wrong.
    Without the glitches I'd certainly had just trashed the plugin, feeling fooled by my own memory. (there were no error messages)

    It doesn't matter if the data is messed by hardware transfer (that specific case) or by a failed program routine on parameters or whatever conditions.
    It's software and shit happens ;)

    Bottomline: the DAW is more than it's summing section - it has to keep bussed data integrity under a lot of varying conditions.
    Interfaces, drivers, plugins, system processes - that's where it's really happening. In the end it's down to the 'trust your ears' paradigm once again...

  • @Telefunky The baseline of what you wrote is, that shit happens and that we need to accept small differences. Right?

    I could live with that, if i knew that a bit of silence was added at the beginning or at the end, so that i could crop the right thing.

    As much as i dislike that BeatMaker does not know how long the track is, at least it can cut cleanly after the given number of bars!

    I will try to use this, to determine what is the right way to crop the Cubasis result.
    And then I repeat the same for Auria, which seems to be a bit too short.

    I need this to be able to compare rendering results, as this would be done by comparing the Spectrum graphically.

  • @tja said:
    I created a new, simple MIDI at 120 BPM and 1 bar, the. froze it (that included the trail) and then made a mixdown.

    Both resulting files are slightly over 2 seconds:

    This looks like what @samu suggested way earlier in the thread. He suggested it is the natural decay of the voice you are using. You misunderstood him, thinking he meant FX tail.

    If you look at the first three notes, you can see that the decay gets cut off by the following note. The final note gets the full decay, since there is nothing saying to chop it off. Why would you want to chop it off, except to get a perfect loop?

    Try this: do the same test, but with a fifth note. Use regions to only render the first four notes. That should chop the decay of the fourth note, and give you a correct loop length.
    Or try with a click sound instead of one with long decay. You may get a shorter loop. If so, I suggest you use the render region option, as well as trim if still necessary.

  • @CracklePot said:

    This looks like what @samu suggested way earlier in the thread. He suggested it is the natural decay of the voice you are using. You misunderstood him, thinking he meant FX tail.

    Yes, decaying notes was what I meant.
    Some synths do not always properly decay to 0 causing longish files.

    If you look at the first three notes, you can see that the decay gets cut off by the following note. The final note gets the full decay, since there is nothing saying to chop it off. Why would you want to chop it off, except to get a perfect loop?

    And it won't be a 'perfect loop' unless the excess is mixed into the beginning of the file :)
    (This is one feature I love in LogicPro X, it creates 'perfect loops' by mixing in the decaying notes, in practice it saves the 'second pass' of a loop).

  • edited July 2018

    @tja said:
    @Telefunky The baseline of what you wrote is, that shit happens and that we need to accept small differences. Right?

    not exactly - the usual argument is that all math is created equal and as such all DAW summing engines have to be identical.
    Even if this would be absolutely true, it ignores the fact that processing input data (starting with conversion of the input file) runs through a multitude of stages that are hard to trace and which may influence intermediate data differently under varying load conditions (or by software flaws that only occur under specific circumstances).

    The (imho) false general assumption is that software performs flawless.
    Digital audio is very tolerant in this context (compared to an accounting app) as it deals with 'analog' data which has analog output as it's target.

    Even if 2 waveforms look different, they may sound absolutely identical.
    You do NOT 'hear' those sample points, but the result of a reconstruction function which is applied after that stage.

    It's same with (white) noise: every snippet (of same length) taken from it sounds identical, but they all have different content ;)

  • @Telefunky OK, i understand better now. Still i will provide some graphical data later ;)

  • @CracklePot the trail you can see was from the frozen track, not from the mixdown.
    Also, both variants get automatically cut at the same length by the mixdown.
    And both are a bit too long!

    I tested this will a sample from BeatMaker 3 with interesting results - i will post later.

  • @Samu I was not interested in the trail, as i just wanted samples of the same length from all DAWs. So, i set the mixdown to cut the trails! The result should have been exactly 2 seconds -and this did not happen!

    I have an interesting result from the comparison that i will post a bit later.

    That feature from LogicPro X sound quite interesting!

  • @CracklePot said:

    @tja said:
    I created a new, simple MIDI at 120 BPM and 1 bar, the. froze it (that included the trail) and then made a mixdown.

    Both resulting files are slightly over 2 seconds:

    This looks like what @samu suggested way earlier in the thread. He suggested it is the natural decay of the voice you are using. You misunderstood him, thinking he meant FX tail.

    If you look at the first three notes, you can see that the decay gets cut off by the following note. The final note gets the full decay, since there is nothing saying to chop it off. Why would you want to chop it off, except to get a perfect loop?

    Try this: do the same test, but with a fifth note. Use regions to only render the first four notes. That should chop the decay of the fourth note, and give you a correct loop length.
    Or try with a click sound instead of one with long decay. You may get a shorter loop. If so, I suggest you use the render region option, as well as trim if still necessary.

    In this scenarios I would also do what @Samu suggested, record the loop for twice as long, and then trim the second half to suit.

  • @tja you said earlier that you wanted to compare the spectra of the rendered output... if that's what you really meant, then sample - accuracy doesn't matter anyway as the FFT doesn't really care about phase...

  • @SevenSystems said:
    @tja you said earlier that you wanted to compare the spectra of the rendered output... if that's what you really meant, then sample - accuracy doesn't matter anyway as the FFT doesn't really care about phase...

    The length of the samples need to be the same, otherwise you cannot compare graphically. Which was the sole reason to post about this non-standard size of Cubasis.

  • OK, i did this:

    Created a 1 bar MIDI in BeatMaker 3 and exported exactly this bar, 2 seconds:

    In AudioShare, you can see the exact 2 seconds:

Sign In or Register to comment.