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.

Removing heavy traffic noise from audio in IOS?

I took a small side job editing a video podcast interview. The interview was conducted outside and there is a lot of ambient noise and frequent traffic noises which are not subtle. I don’t expect to be able to remove all of this noise but I’m hopeful that I can find a way to make it less distracting. There’s an app, Moises, that uses AI to isolate music tracks into separate vocals, drums, etc. Would this work with a amateur raw audio that has no panning?

I found some youtube vids about Klevgrand Brusfi and it looks like it does a good job of removing ambience, but am thinking the AI of something like Moises may actually be better for very intrusive noises.

Any suggestions?

Comments

  • edited November 24

    Brusfri is good for ambient sounds that you can easily isolate. Think constant highway noise off in the distance. Would be ideal for the ambient noise in your recording, but I don't think it would be nearly as good for heavy traffic noise. It's worth trying especially at the sales price. I don't imagine a noise gate will work if the traffic volume is heavy enough, because it will go over the threshold and there is also the gating effect to contend with. Brusfri is pretty close to a gate but not exactly the same. Sometimes one or the other works better in my experience (or both in serial).

    I don't know about Moises but I do know that applications of that type tend to introduce non-subtle artifacts that I find rather unlistenable. If you use it to remix, the artifacts blend into the mix and become a lot less noticeable. YMMV

    The best affordable tool for this job would actually be desktop-based and that is Waves Clarity Vx. I have not encountered anything as useful for such noise at anywhere near that price point. If you can do the work on a desktop system, you will get far better results and you could probably get it done using a trial version. It also can introduce artifacts, but combined with a gate or brusfri and keeping it subtle, you will get a usable result.

    My suggestion for the best result is don't do it on iOS. If you are limited to iOS then Brusfri may be your best option.

    [EDIT: actually your other option is a multiband expander. I'm only familiar with Fabfilter Pro-MB but there may be others. You can google generally how to use one, but here is a starting point: https://sound.stackexchange.com/questions/7034/how-to-minimizing-noise-and-reverb-with-diy-multi-band-expansion ]

  • Another vote for Brusfri

  • Send a sample and we can try brusfri on it, be sure to add some pure noise

    If you really need it done use the nVidia thing rtx voice

  • Also krips seem to say its free and better than rtx: https://krisp.ai/blog/nvidia-rtx-voice-krisp/

  • I have Brusfri and it’s very good for constant noise (such as microphone hiss) it can listen to, learn and then know what to ignore. I don’t think it’ll prove very successful with traffic sounds which are unpredictable and varied though.

    You can get pretty good results with Dolby On too. They do know a thing or two about noise reduction after all. It’s free so you may as well try it out. It gives you (slightly) more in-depth controls than might be initially obvious, if I remember correctly, so do dig around.

  • edited November 24

    Looks like 4pockets has a multiband compressor/expander for iOS and a lot cheaper than Fabfilter's, but only 4 bands vs 6.

    If you are limited to iOS, and budget limited, I would recommend a combination of brusfri and the 4pockets expander and a lot of time to dial in the best results. Should get you something usable.

    Here's a good tutorial I have bookmarked on using Waves C4 for this type of task.
    https://web.archive.org/web/20130602013740/http://www.stavrosound.com/blog/wordpress/2011/07/dialogue-clean-out-the-noise/

    If you have the option, use the classic film dialogue technique of adding sounds (typically music) over bits of noise that you just cannot remove using these other tools.

  • @Robin2 said:
    I have Brusfri and it’s very good for constant noise (such as microphone hiss) it can listen to, learn and then know what to ignore. I don’t think it’ll prove very successful with traffic sounds which are unpredictable and varied though.

    You can get pretty good results with Dolby On too. They do know a thing or two about noise reduction after all. It’s free so you may as well try it out. It gives you (slightly) more in-depth controls than might be initially obvious, if I remember correctly, so do dig around.

    Dobly on seems nice! can you import audio files?

  • Haven't used it myself, but you could try this: https://audo.ai/noise-removal

  • @NeuM said:
    Haven't used it myself, but you could try this: https://audo.ai/noise-removal

    the examples are mindblowing

  • @cokomairena said:

    @Robin2 said:
    I have Brusfri and it’s very good for constant noise (such as microphone hiss) it can listen to, learn and then know what to ignore. I don’t think it’ll prove very successful with traffic sounds which are unpredictable and varied though.

    You can get pretty good results with Dolby On too. They do know a thing or two about noise reduction after all. It’s free so you may as well try it out. It gives you (slightly) more in-depth controls than might be initially obvious, if I remember correctly, so do dig around.

    Dobly on seems nice! can you import audio files?

    Yeah. Click on the white ‘quaver note' button and you’ll see the import button on the screen which opens. Allows video import too (i think?).

  • edited November 24

    @cokomairena said:

    @NeuM said:
    Haven't used it myself, but you could try this: https://audo.ai/noise-removal

    the examples are mindblowing

    Plus, it's free for up to 20 minutes of audio per month. I have a feeling that if they get more (free) uploads it has the net effect of improving their machine learning algorithm. That's why they're willing to "give away" the service for now.

  • edited November 24

    Thanks for the tips. I’ll give these options a try in the next few days and report back.

    Also, I don’t have a PC. IOS only.

  • A tip for the future. Buy a microphone foam shield and put it on your phone (if you're using a phone, I hope lol)
    It will act as a good shield for some outside noises (mainly wind)

    That with a noise gate while recording will probably help a bit. Now to cleanup what's recorded, besides the tips above I would recommend some surgical EQ and spectrum editing (but that I think we cannot do on iOS)

  • @NeuM said:
    Haven't used it myself, but you could try this: https://audo.ai/noise-removal

    I tried this with .m4a and .mp4 and it says “encoding failed, file type not supported” even though the documentation says both of those are supported.

    I’m trying this from Chrome browser on my iPad.

    It’s disappointing because the examples are stellar.

  • edited November 27

    @Sabicas said:

    @NeuM said:
    Haven't used it myself, but you could try this: https://audo.ai/noise-removal

    I tried this with .m4a and .mp4 and it says “encoding failed, file type not supported” even though the documentation says both of those are supported.

    I’m trying this from Chrome browser on my iPad.

    It’s disappointing because the examples are stellar.

    Hmm. That's strange.

    Have you tried uploading other formats? ".wav" for example?

  • Moises does an amazing job! It only does 5 minutes for free, but I’m willing to piece two clips together to make it work. It works as well as the audo.ai examples.

  • edited November 27

    And you listen with a good set of headphones and don't hear any artifacts? I would not call the Bowie or Marvin Gaye separations in these examples very good, but others apparently think they are acceptable. I absolutely hate autotune, and it definitely hits the same part of my brain. There's something going on with the pitch for example.

    https://www.theverge.com/2019/11/5/20949338/vocal-isolation-ai-machine-learning-deezer-spleeter-automated-open-source-tensorflow

  • Yeah> @kidslow said:

    And you listen with a good set of headphones and don't hear any artifacts? I would not call the Bowie or Marvin Gaye separations in these examples very good, but others apparently think they are acceptable. I absolutely hate autotune, and it definitely hits the same part of my brain. There's something going on with the pitch for example.

    https://www.theverge.com/2019/11/5/20949338/vocal-isolation-ai-machine-learning-deezer-spleeter-automated-open-source-tensorflow

    Right, it’s not perfect. I can easily hear the artificiality. But this is a recording of an interview made with a phone and no external mic, outside on a busy street. I can’t wrap my head around how beyond amateur the raw recording is.

    So, a filtered voice that can now be comprehended is 100 times better than what I was given. Also, most people will hear this interview while listening on their phone, not with a decent set of headphones.

  • edited November 27

    @Sabicas said:
    So, a filtered voice that can now be comprehended is 100 times better than what I was given. Also, most people will hear this interview while listening on their phone, not with a decent set of headphones.

    Indeed it is 100x better. It's also true that audiences have become a lot more forgiving of lower quality audio in the past decade, because the bar has been lowered by amateur production techniques (and crappy bluetooth headphones). As long as it is good enough for you and the client, my perfectionism doesn't matter. :smile:

Sign In or Register to comment.