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 StoreAudiobus is the app that makes the rest of your setup better.
Comments
@Blue_Mangoo thank you for these and your information in the other posts - great explanations of all this stuff and I've learned loads!
Glad to hear it is useful to someone.
@Blue_Mangoo
Thanks for your videos.
I'm not an expert on the topic, imo oversampling (specially on iOS) should/could be a rendering option, where plugin depending on its function/requirement would do it offline.
Your video shows perfectly well that it's all a very delicate balancing act, which means depending on use, user should be able to pick.
Just out of curiosity, I would like to know/see what happens with that ringing after downsampling. At the end of the day it's not really about "can you hear" over 20k, more about when you mixing 20-30 channels how much aliasing can you tolerate / how much "headroom" you have on top of alias free freq range.
Thanks again! 👍
Useful to many.
I totally agree!
Would it be possible to just use a low pass set at the most “frequent” aliasing offender to rectify this potential headroom situation?
Really not an expert, just thinking out loud...
I think, there is no "most frequent" with fx like comp, saturation, distortion.
Just to clarify in case anyone is confused on this point:
All demos in this video are upsampled AND downsampled back to original sample rate.
Now if I am understanding your question correctly, I think you are wondering what would happen if I further downsample again to remove the top frequencies, because it looks like the trouble is near he top of the spectrum so if I downsample again then I should be able to cut that problem area out and have a clean signal.
I am sorry I didn’t do this on the video because the result is very interesting to see, and a little frustrating. Basically, the answer is no, this doesn’t help at all. That ringing at the nyquist frequency is caused by the lowpass filter we use for antialiasing. Unfortunately, it can be traded for other problems but it can not be eliminated. Here is what happened when I tried to eliminate the ringing by filtering it out, which is what would need to happen before I could safely downsample:
Conclusion: there are three artefacts of filtering that may occur whenever you design a filter for downsampling or upsampling. They are: delay, ringing, and loss of high frequencies that you didn’t intend to loose. You can nearly eliminate any two of these problems at the expense of making the third one significantly worse but you can not eliminate all three at the same time. Any antialiasing filter you can build will be a compromise between the three.
@Blue_Mangoo brilliant video ! Thanks for sharing this information.. wondering that you have just 415 followers on YT ! Man you deserve at least 415 000 ! Great educatio stuff, just keep it goimg !
I am trying to wrap my head around this idea. Not sure if I got it right yet. Perhaps the answer in my previous post already addressed it, because the fact remains that there are three artefacts of filtering that can be traded off but not all three can be eliminated simultaneously.
I thing that by “ lowpass filter out the most frequent offender” you mean that I filter out everything above a certain point, Placing the filter cutoff at a point where I get to keep most of the sound I want to keep and loosing some sound I am willing to give up then the answer is definitely yes. This works well if you are making electric guitar effects because electric guitars don’t have much useful content above 10 KHz so you can safely put the filter cutoff there.
However, this solution can not be used for plugins designed for general audio content because in the general case we don’t know which frequencies are important and we have to assume that any audible frequency (20 Hz to 20 KHz) needs to be preserved.
Thanks.
We started doing this type of video just a few months back. I also hope to see more subscribers as we continue. This topic is obviously way too technical for a general audience but I personally wish that there were more people sharing this level of technical detail on The internet because I need to learn this stuff in order to do my job.
Thanks for another awesome video! I’m a bit of a noob at a lot of this stuff and tour videos have really been helping me understand a lot about the studio side of things
Ah yes thank you. I was referring to being okay with losing some of the frequencies. Guitar guy in a guitar world.
Thank you for this suggestion. I did not know that audiophile DACs have that feature. I still listen to music with my iPhone.
Yeah, for guitars there is room to distort the top end, then just filter it out later.
Just wait 10 or 15 years, and you won't have to worry about frequencies above 15k.
Ok I see what they are doing. The FIR filters are... FIR filters. The Short Delay filters are IIR filters, possibly similar to what I was using in the video. The SDLY SHARP is like what I was getting in the video when I set the transition bandwidth down to 2.5 percent of the spectrum and the sdly slow is probably more like what I had when I set it to 10 percent. I don't currently have code handy to do a demo of the FIR filters.
Yeah, I'm afraid of that. Maybe I'll have to get a new job when that happens.
Naw. You’ll have additional wisdom and know-how and young interns to alert you to high-end artifacts.
@Blue_Mangoo
i have one question... which part of oversampling process adds that additional content to initial simple response, or at least which one is more responsible for that "tail" which is added ? It's upsampling part or downsampling part ?
I mean, i would like to see how this wave looks also in "upsampled" stage ..
It looks like it comes equally from both the upsampler and the downsampler. I tried adjusting just one of the two so it doesn't ring like that, but the ringing still persists in the output until you fix both of them.
I have an idea about how to eliminate most of the ringing, in exchange for some extra samples of delay at the beginning of the impulse response, by using a series of critically damped lowpass filters. Haven't tried it yet. If it works, I will post a video of it tomorrow. Lately I am finding critically damped filters to be extremely useful for antialiasing. They are not very powerful in terms of filter cutoff slope and probably for that reason they are rarely used in audio applications. But they are the strongest lowpass filter type that has absolutely no ringing.
this exactly came to my mind, that why i was asking .. interesting stuff, looking forward to next video
Before doing the previous video I tried all sorts of lowpass filters, to no effect. But I didn't try the critically damped ones yet because izotope RX EQ doesn't have an easy way to set up a cascade of many of them in a row, which is what it would take to clean up that ringing mess. The main problem with critically damped lowpass filters is that their cutoff slope is so gentle that you need a large number of them to get a sharp cutoff and stacking up many of them in series could cause a lot of added delay.
I'm not that deep into dsp algos and math behind individual filter implementations, but what about butterworth, from my brief sketchy understanding, it should be very close to critically damped ones, but has more sharp cutoff ??
Butterworth filters have ringing near the cutoff frequency. It's the ringing that allows it to push the filter curve up at the cutoff and achieve a flat frequency response in the passband. If you need them to not ring, the critically damped filter is the one to use. It's kind of fundamental in the sense that the simple first order lowpass is critically damped, and if you cascade any number of identical first order filters to get a higher order filter, the higher order filter that you get will also be critically damped.