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 : I was thinking of this in your straight parametric EQ rather than the amp modeler though it would be great in the amp modeler. Maybe an IAP available at a later date?
I mentioned this because all this discussion has made me interested in systematically comparing the output from the my various guitars and better understanding the characteristics of the sounds I like.
We tried being secretive about what we are working on to avoid this. Giving out information about something that isn’t built yet also means that you are going to show people all the wrong turns you make. much of what we show in our videos will get scrapped later on and never make it into an app. Basically we’re afraid to let people see us mess up and to let them see how many dead ends we reach before releasing a product. But one day I was working on something interesting and I thought, “People would probably like to see this.” And I figured, does it really matter if we make mistakes get people excited about new products only to let them down when some of our projects fail? And do we really mind if they request many features that we are not capable of delivering? Why not just let everyone enjoy the chaos and we will enjoy it along with them.
I see what you mean. I think we could do that, but honestly fab filter has done such a great job, I don’t want to compete with them until I have an idea about how to improve on what they did.
This feels like the audio engineering lab with a lecture and Q&A.
The information applies to mastering and general audio work.
Even simple tips like:
using multiple copies of the EQ for pre/post processing
IR reverbs are also EQ's based on models from recordings
tubes are wave shapers
clipping is the shortest path to an (almost perfect) square wave
Keep the video lectures coming.
That's your most reasonable sentence in the whole discussion
Your sound examinations are of course valid attempts to shape sound - as that's how it was done for ages before convolution modelling appeared on the scene.
But this is not 'amp modelling', because amp stages have a dynamic change of signal content during it's path - something you already experienced yourself.
Do you really expect to find a new shortcut with 'tools' that have been applied for decades ?
Such experiments are cool because they help those with less hands on experience to get a good idea about how sound is influenced - and it's often enough to get a tone satisfying in a certain context.
There is no need for cabinet/IR/whatever modelling - record a distortion pedal, a real amp or just send the raw guitar sound through some waveshaper of the brutal kind.
Add some reverb/delay to fake an acoustic environment (or leave it out) and that's about it. The track or band context defines the fun.
On the other hand a lot of people want 'real amp' behaviour of their modelling plugins.
It took a long time for Line 6 from their first Pro Tools plugin (tinny crap) to the Helix pedal.
Which also applies to S-Gear (Scuffham), Positive Grid, only in a shorter timeframe.
But (imho) it's bizzare to challenge such apps with eq, saturation, waveshaping, etc applied in almost arbitrary fashion.
Obviously it takes more - otherwise someone would have found the solution around 1995 and all those modelling amps never had happened.
Sidenote regarding Kemper: atm it's the holy grail - but not because it sounds great...
It allows to 'copy' the sound of famous, rare, expensive gear and that's the argument within the guitar scene, as @jacou experienced:
All kind of sound experiments and ways to build special setups are great, but 'amp modelling' simply is the wrong (and misleading) label.
(sorry if this comes across like an a***hole post - but it's written with best intentions)
In short, yes.
There are obviously a few hardware manufacturers who make amp models that are indistinguishable from the real thing in a blindfold test. I have not seen any plugins for desktop computer or iOS that can claim that. It’s still an open problem.
The key idea that motivates me is an observation about the distinction between Linear and Non-linear modelling.
Resistors andcapacitors are usually simulated with linear models. Tubes, clipping diodes, and power supplies have to be modelled using non-linear methods.
A Linear system can model EQ and volume boost /cut but it does not model clipping or saturation of any kind. Effects like saturation, clipping and sag are classified as non-linear.
The current state of the art circuit modelling techniques allow us to quickly and easily look at a circuit diagram and make a LINEAR model of each individual component in the amp. Unfortunately a linear circuit model can only do what a parametric EQ does: boost or cut certain frequency ranges.
Unfortunately, the technology for modeling the nonlinear components of the circuit lags behind the linear models. It is less accurate and in general we don’t know the best way to simulate it.
So here is why I am excited about doing things the way I am doing them:
I’m not satisfied with the way most software amp models sound. The problem with them is that they have detailed linear models of the preamp EQ and other filters combined with crappy nonlinear models of the tubes and power supply. My plan is to throw away the linear models of the preamp EQ because a parametric EQ does exactly the same thing except that it is way more powerful and customisable and many audio enthusiasts know very well how to use it. I have always wanted to replace the bass,mid,treble knobs on my amp with a parametric Equalizer; now my dream can come true.
Now only one thing remains for me to do: make a good nonlinear model of the tubes and power supply.
Here is what’s new about this approach:
Instead of coupling a good model of the amp EQ with a crappy model of the tubes and power supply, I want to couple a good model of the tubes and power supply with a parametric EQ.
If you compare between amplifier brands, you find that many of them are circuit clones of each other except for a few resistor values changes in the preamp. It’s sad that in a software amp app the user interface for switching between those two amp tones is selecting a picture of the amp from a list. Wouldn’t it be better if we could see a graph of the preamp EQ curve and just pull the mid scoop where we want it or boost the bass and cut the treble to taste?
Well if my tube stage saturation turns out to be crap then we will be back to 1995 with this app.
But if a few iOS guitarists can start making their own preamps using parametric EQ then even if it’s coupled to a 1995 era saturator, we still have something more flexible than what was available back then.
Ok I’m off to make sure the tube distortion sounds better than it did in 1995.
I recently got excited by the demo's for the new Amplitube "Brian May Collection" and as I played through the preset I kept thinking: the AUv3's in AUM sound better than this so I think deconstructing the Amp Sim functionality into a collection of user selectable AUv3's in the best option for many IOS centric guitarists.
There will always be hardware with better audio processing features but it's nice to get some sonic wonder in a pair of headphones on a truly mobile device with reasonable battery life.
I was thinking about this yesterday and wondering if I haven’t been too hasty in my conclusion that most software amp sims aren’t modeling the tubes and power supply well. First of all, I have not actually tried them all. I tried a few of the most heavily marketed ones several years back and didn’t like them but it seems like the ones most people rave about are ones I haven’t tried. I keep hearing that scuffham is awesome but still haven’t tried it out.
Last night I watched an Anderton’s TV interview with Christoph Kemper and learned some interesting things from it.
Christoph Kemper made the point that many guitarists listen to software models and immediately say “it doesn’t sound real” because they aren’t used to hearing their guitar amp through a microphone. The problem there is that when we record the IR of the speaker cabinet, what we actually get is the IR of the speaker cabinet + the microphone + the space around the cab. And if we play that through speakers, what we are hearing is the following signal path:
IRCab > IRmic > IRRoomEcho > MonitorSpeakerCab > RealRoomEcho
When what we are used to hearing is just this,
RealSpeakerCab > RealRoomEcho
Obviously the first doesn’t sound like the second because it goes through a mic and the second does not. Even if the mic is perfectly modelled, it’s still never gonna sound like the real guitar amp not recorded through a mic.
This is where you really come out ahead if you can find a way to model the IR by ear: using your ear alone allows you to eliminate the mic from the IR.
As I think on it, I am still convinced that all the amp sims I own have bad tube and power supply models, but there are probably others that model the nonlinear stuff very well.
regardless of whether they are accurate or not, the tone we are getting with parametric EQ sounds exactly like WE want it to sound, and since I started doing this, I also feel like I am finally getting the amp tones I have been wanting all these years.
@Blue_Mangoo : this is why I have maintained in other threads that one has to compare the amp-simed signal to playback of one's amp recorded. The experience of being in the room with the amp -- even with the best recordings is often different from having been in the room when the recording was made.
That’s exactly what Kemper said in the interview. But most guitarists want the model to sound like the amp, not a recording of the amp. That’s why the Kemper amp has a knob on it for eq’ing out the mic sound.
Sounds pretty cool!
Thank you
no other hardware involved. I’ll try and post more audio/video soon.
Yeah after getting a feeling for that method it’s pretty liberating. No more endless amp sim/IR searches + adjustments to get a relatively good compromise. It’s like working with EQ presets instead of using the eq directly.
I get pretty nice results with using steep high cuts in the post saturation eq. Boosting the lower high end before the saturation and cutting off the high frequencies in the post saturation EQ. That’s why I’m using the Pro-Q to get the steepness of the high cut (don’t know the technical term for that steepness).
Any chance that can be added to your EQ? (Might be quite complicated though)
The only rule of EQ is if “it sounds good do it”.
However, analog hardware almost never has steep filter cutoffs. One reason for that is that steep filters are much more complicated and expensive to build.
I am saving a large library of setups in AUM for various different guitar amp tones. Because we recently updated our EQ to include filter shapes with less steep slopes, I went through my library of EQ presets and replaced all the 12db/octave high cut and low cut filters with the more gently sloped 6db versions. In the presets I examined today there was not one case where the switch to a gentler slope did not improve the sound.
There is a time and a place for steep slope filters and I’m glad that fabfilter has them. I don’t plan to put them in our EQ though because I want our EQ to stay focused on being really fast to dial in the sound, and simplicity is essential for that. In 99% of EQ situations, steeper filters get you into trouble, meaning that while you are fixing one thing you make something else worse. For the remaining 1% there’s fabfilter.
Thanks for your response! Yeah I agree I usually also use soft eq shapes. Nothing too radical. But with that amp EQ it weirdly works well with a very steep slope. I use something like 48db for the high cut.
My try to explain that (without being a very technical person) is that guitar amp speaker cones are usually pretty big so they can’t produce any high end (not even a bit I guess). So in my logic that would be a rather steep high cut slope. Would be interesting to dive more into the physics of that.
What are your thoughts on that?
I also don’t try to reproduce real amp cabs but I’d like to get some of that “real amp” feel.
I agree. Most of the time the gentle slope works best but there are situations where something near 48db is appropriate.
I’m also not sure how steeply the speakers roll off the high end above 4000 Hz.
When I need to cut steeply in the high end of the speaker cabinet EQ I do it with a pair of bell filters rather than a high cut. The reason is that I find cutting the high frequencies right down to zero doesn’t sound right; it lacks presence and clarity. We want to reduce them but not eliminate them.
To do a steep cut at 3KHz without totally obliterating the high end, try boosting at 250Hz with a Q of 0.35 and pulling the skirt of that bell filter down -10 dB to cut high end out. That will give you a slope that’s way too gentle. To make it steeper, take a second bell filter with Q of about 1 and boost back up at 2100Hz. The combination of those two will make a steep cutoff in the high end that doesn’t actually go to zero:
That’s close to the slope of a 48dB high cut but without the total loss of sound near 24Khz. If you need it steeper than that you can use three bell filters, the lowest one with the widest Q and the highest one with the narrowest Q.
Could you please include the frequency and dB scale in the screen shot?
Another comfy feature in bell filters is adjustable slope. Makes life a bit easier
These screenshots show the filter info, but the 10 dB cut in skirt gain on the lower filter isn’t printed there:
What did you mean by “adjustable slope”? Is that different from Q?
Thats a good workaround!
I'll try that on my phone where there is no Pro-Q
Aaah perfect
No, same thing, call it Q, call it slope, call it steepness...
All of the filters in this app have adjustable Q or slope.
Oh, even better then!
i'm looking forward to the release of your saturator.
i've explored this technique a lot on laptops in the past, though not since 1995 as @telelfunky has ;-) i've also explored it quite a bit on ios. fun.
are you considering allowing for the bypassing of each of the features (eq, spatialisation, etc)? it would be nice if the user could use resonant multi-pole filters, or play with polarity and phase pre/post the saturation stage.
you also mention sag. are you implementing a sag emulation?
will the saturator be capable of interacting/responding to pre gain/pre distortion/pre overdrive, ect. auv3's? though i've never built a tube amp, i've always noticed how preamp pedals add to the responsiveness of tube amps (gain staging). i recall dumble (and others) utilising a FET somewhere in (some) of their circuits to affect responsiveness and distortion character.
have you explored power amp/magnetics ?
for those that are already playing around with these ideas; try a little pre and post compression. speakers compress. for those that are wanting some responsiveness; i've found that low latency helps a lot with responsiveness. also, playing through speakers (moving air/ feedback) is important.
unfortunately ios doesn't have input latency lower than 64 samples. maybe this will change in the future? i don't know how much latency is contributed by audio processors/effects on IOS. do audio processors on IOS have latency compensation? are their host that accept latency reporting? do you know @Blue_Mangoo?
Yes.
Yes.
We have one model for sag with adjustable parameters. So far I have only tried one stage of sag in the model. It seems to work for modeling the big sag in the power amp or the little sag in the preamp but the power amp sag sounds too compressed when it runs without a preamp, meaning that it accentuates the pick sound at the beginning of each note more than a real amp would. So I plan to test two stage model tomorrow, hoping that with the fast compression from the preamp stage squashing down the attacks, the power amp stage sag will be able to compress without such a punchy attack.
To directly answer your question, we are not really looking at this from a deep understanding of electromagnetic effects in power supply transformers. The model we have comes from looking at oscilloscope graphs of input and output in amplifiers and making a software model that has a similar looking output without doing detailed calculation about how the amp does what it does. So I did not even consider the concept of magnetism when coding this, but it sounds good and the oscilloscope output looks correct.
Low latency helps because it reduces the time between when you touch the strings until when you hear the sound, which is hugely important for the feeling you have as a musician when playing through an amp. If you don't have the right feel, you won't play as well, and the sound will therefore suffer. However, if you were to turn off your speakers and record your guitar using low-latency, then process the recording using low-latency, it would not sound different from what you get if you record using high-latency, as long as you keep your speakers switched off throughout the recording process. So the effect of latency is mostly psychological, but still extremely important. The only way latency could actually affect the sound directly is through the feedback that happens when the sound from your speakers vibrates your strings.
I am not aware of any iOS host app that supports latency compensation. Based on what I wrote above, even if they did compensate, it wouldn't make your guitar sound any better because latency compensation doesn't eliminate latency in realtime. All it does is keep the play head aligned properly when you play back a recording.
Just interjecting here to say that this thread is most excellent. Please keep mining this deep seam of knowledge and experimentation. 👍
I'm definitely a fan of the less is more approach. On one hand to save CPU/latency and on the other hand to keep things simple/straight forward.
I assume that virtual mic placements are also just changes of EQ filter curves in the background.
About the dynamic/compression. It's incredible how much you can achieve compression through the saturation.
If I increase the pre sat gain and then cut off the high frequencies (where most of the distortion is audible) the signal still sounds relatively clean but has a great compressed/round amp feeling to it.
Through the higher saturation the frequency band becomes quite rich so that the guitar doesn't sound very honky or unbalanced anymore. I hope you get what I'm trying to say
I probably should upload some examples of it..
in the end all that matters is the sound. your answers about sag and gain stage responsiveness are exciting!
yes. it's noticeable when i play faster and with a cleaner tone. when using delay/echo and/or reverb it's less bothersome.
i understand. i was babbling. was mostly wondering how much latency each effect instance might contribute. i'll answer that question myself, with more ios exploration. you would only know the latency details of blue mangoo plugins. i'll pose my question about lower input latency to Apple, as i should.
thank you for your videos, plugins, and your time on the forum.
can't wait for the new waveshaper.
Latency is the biggest complaint I hear from real guitar players when I give them an IOS amp simulator to play with. Remove latency and deliver a good tone and it might get more players to value and IOS Amp Sim.
Latency is a hard problem for iOS developers because it depends on the operating system more than the app. iOS latency seems pretty good to me with a buffer size of 64 samples though. At 128 I feel the lag but don’t mind it much. At 256 it noticeably affects my playing and above that unusable.
I found this latency meter app:
https://apps.apple.com/us/app/round-trip-latency-meter/id1427507645
I intend to try it out. It’s good to measure these things to make sure we aren’t fooling ourselves into being more or less picky than what’s reasonable.
If you stand three metres away from a real analog guitar amp there is 8.7 milliseconds of latency due to the time it takes for sound to travel from the amp to your ear.
3m / 343 mps = 0.0087
I use that as a benchmark because I never felt concerned about latency when using a 10 foot long guitar cable.
Just a little tipp for anyone who already has an app that can record audio and show the time scale in milliseconds (I'm using Twisted Wave for such measurements):
Use a short spike for measurements, maybe a very short & tiny guitar pick noise and record both the input and the output signal. It's usually no problem identifying and measuring these on the timeline. I always use multiple such transients to make sure to reduce the error involved with measurements done by the eye.
For measuring audio latency from hitting a key to hearing sound, I put large headphones on top of the keyboard, the iPhone in between and just record a very hard keyboard hit plus the audio coming from the phones. Sounds weird but works well for several reasons:
The distance between sound source, "speakers" and microphone is minimal, and using headphones, you avoid additional latency produced by the iOS speaker audio processing (like volume-dependent EQ).
Not as elegant as a fancy "round trip time meter" but it works.
The latency measurements shown above are from my iPhone 6s. At 48khz, not so good, but I’m very pleased with the results at 96kHz sample rate