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.

Spleeter

Separating stems from a track: new progress based on AI and deep learning was recently released. See this video for some good and some not-so-good examples. Overall quite impressive!

Comments

  • It is impressive, imperfect but definitely full of possibilities. This popped up a couple of weeks ago and there is already a simplified web version. MP3 only but a great way to dip your toe.
    https://melody.ml/

  • Been using it a lot the last weeks. Have had some crap results and also some pristine results that are like stems straight from the label. What a time to be alive.

  • Is this on iOS & Mac? Thanks in advance for info

  • You can probably use that melody.ml link from iOS. The git version requires a lot of dependencies that are OSX.

  • @bounce said:
    You can probably use that melody.ml link from iOS. The git version requires a lot of dependencies that are OSX.

    @bounce cheers Much prefer if was an iOS app when there’s no internet available.

  • edited November 2019

    Here's another implementation of the Deezer machine learning open API.

    https://moises.ai/

    I'm currently comparing to see if there are (m)any differences between this and http://melody.ml

  • Is it MP3 only like melody.ml?

  • I think the real beauty of the spleeter code is that you can feed it a library of files to learn a style of music. This would require access to multitrack stems, though.

  • edited November 2019

    @RJB said:
    Is it MP3 only like melody.ml?

    It works with WAV as well, but does require signup via email address.

    So far, I can’t tell much of a difference in quality in comparing MP3 files, but Moises.ai gets the edge due to the higher quality potential of using WAVs.

  • Yes it is a powerful algorithm, perhaps the most powerful the world has ever known. But can it learn to dubstep?

    THAT is the question before us, ladies and gentlemen.

  • @eyesunderground said:

    @RJB said:
    Is it MP3 only like melody.ml?

    It works with WAV as well, but does require signup via email address.

    So far, I can’t tell much of a difference in quality in comparing MP3 files, but Moises.ai gets the edge due to the higher quality potential of using WAVs.

    Thanks for the update, I think I’ll sign up

  • Quick resurrection. I played with this on Win10 this morning. It is ableton only and requires a bit of command line and system-fu but totally manageable:
    https://github.com/diracdeltas/spleeter4max#spleeter-for-max

  • @kewtips said:
    Quick resurrection. I played with this on Win10 this morning. It is ableton only and requires a bit of command line and system-fu but totally manageable:
    https://github.com/diracdeltas/spleeter4max#spleeter-for-max

    Thank you for that link.. I’ll be checking it out as I’ve just finished updating to Catalina with a fresh install of Ableton 10 on a MacBook Pro that was kindly given to me.. big hearted people.. you gotta love them.. and I do! ❤️

  • edited May 23

    Just installed this on Ableton this morning. Does it need more than 8GB of RAM? I have tried slicing up the audio file into smaller bites but it still doesn't spit anything out.

    On a more optimistic note, first time I've used Max 4 Live's content site. It's like patchstorage.com's files for Drambo etc, but for Ableton. Love it.

  • I have 8GB ram and it worked for me. But I was mainly doing 16 bar loops.

  • You do know there’s an app for that. It’s called “Let’s Unmix” by Hikaru Tsuyumine.
    Requires iOS 13.2 of higher. Is free to start with but needs an iap of 5.49 euro’s.

    Here’s the Appstore description:

    • Separate music into vocals, drums (percussion), bass and other instruments.
    • Separated tracks can be saved and exported to other apps such as DAW.
    • You can control volumes and panning of each track separately for playback
      and export. Playback speed also can be adjusted.

    • You can cut specific regions of the tracks for playback and export by defining
      the area of the waveform view.

    • Source music can be imported from Media Library, other apps or cloud services
      such as iCloud, DropBox , Google Drive and so forth.

    • Supported file format as source music are mp3, m4a and wav. Supported output
      formats are m4a and wav.

    • The optimal sampling rate of the source file is 44.1KHz (CD quality). Sampling rate
      of output file is always 44.1 KHz regardless of the sampling rate of the source file.

    The iOS Community did 2 video’s on it.

    The First one rather short and a little unclear at the beginning.

    The second one has his famous pointing finger in it. Demo start 55 seconds in.

    And ofcourse the Appstore link:

    https://apps.apple.com/fr/app/lets-unmix/id1500367180?l=en

    Haven’t downloaded it myself. I did use the Windows version. But be warned,
    installing it is a hell of a job, but i’ve succeeded thanks to this YouTube video:

    It’s a big install. 5.5 GB if i remember correctly.

  • This is cool, haven’t tried to install on my iMac but I want to, seems like a headache. I saw it recently on a Mr Bill youtube video.

    @Silvertip I wonder if that iOS app is using the spleeter algorithm or whatever. I might just get that app eventually...

  • Just went to his website via the link on the Appstore and seems there are two
    extra apps. One specific for vocals and a second one for drums and percussion.

    Vocals
    https://apps.apple.com/us/app/vocal-extractor/id1438806137?mt=8

    Drums
    https://apps.apple.com/us/app/drums-extractor/id1471602693?mt=8

    He doesn’t say exactly if he is using the spleeter algorithm, but he mentions:

    “Although the quality of vocal separation is varied depending on the source
    data ... we use Deep Learning technologies which allow us to separate vocal
    from monaural recordings or not center pan vocal recordings”

  • edited May 24

    Not iOS but was considering the this on the desktop will definitely have to compare to Let’s Unmix app.
    The remix feature is also based on Spleeter algorithm. Starts @ 00:39

  • @joniqwest said:
    Not iOS but was considering the this on the desktop will definitely have to compare to Let’s Unmix app.
    The remix feature is also based on Spleeter algorithm. Starts @ 00:39

    Oh nice, there is a 30 day trial. That company has a sweet free desktop Chorus plugin (Multiply)

  • The Let’s Unmix is using the same tech as the Spleeter program, if not the open source code itself it’s using the Open Unmix algorithm that’s said to be very close. Either way you won’t (for now) get multitrack session results, there ARE artifacts. But what is done is amazing.

    I bought the iOS version Let’s Unmix, it’s free and you get 90 seconds of a track, same sound quality, you just can’t save it. The full version IAP was $5 and it’s worth it.

    Vocals and drums, usually center panned, see to sound best. Guitars and keys, background elements are a mixed bag. The bass was the big shortcoming, in all the songs I tried (some Zeppelin, The Police & some old demos I did on PortaStudio) the bass just sounds faint and very underwater.

    To me it's worth it as a musician who loves getting into recordings, the scholarly aspect if you will, but I also wouldn't say no to sampling a drum element from something, like a cool snare. Remixers and hip-hop samplers will absolutely dig this. It's fascinating how far the tech has come. These algorithms that brought us auto tune and this were unthinkable years ago, now digital mix separation is on iOS...heavy.

  • The return of bastard pop is imminent.

Sign In or Register to comment.