A low latency guitar effects processor suitable for running on a Raspberry Pi | Hacker News Reader

A low latency guitar effects processor suitable for running on a Raspberry Pi

github.com

264 pointsAudiophilip142 comments

> Note that since the server program exposes an HTTP server you can actually navigate to the IP address of your PI from any device connected to the same wifi and control your pedal chain from there.

That's a killer feature for me, hiding at the end of the README. I have a Fractal Audio FM3[0] at home, and the only way I edit my patches is using their editing software over a USB connection to the device. Adding the ability to program (and even control) my patches live over any wifi-enabled device is even cooler!

[0] - https://www.fractalaudio.com/fm3/

Always hear about the Fractal Axe FXIII--seems like the gold standard in guitar FX. Didn't realize they had smaller, non-rack-mount form factor devices. Very interested in trying out the FM3 now that you brought it to my attention. Thanks!!

The Mod Dwarf (and its predecessors) also allows web control and it supports standard LV2 plugins :)

https://mod.audio/dwarf/

https://wiki.mod.audio/wiki/LV2

The first site is remarkably brief about its actual capabilities. Is there somewhere with more information about it?

Simplifying a bit: It's a pedalboard with a Linux SBC inside where you can load LV2 plugins and chain and route them as you wish. It has a sleek web interface for management and some short of "pedal store". Like an advanced multieffects pedalboard. Originally the MOD devices where crowdfounded.

There's a small introduction video: https://www.youtube.com/watch?v=5USQ7BsZ054

And Modep (as @tcrenshaw wrote in another comment) this is a MOD emulator for Raspberry pi, just in case you want to play with it: https://blokas.io/modep/

I think the hip new guitar effects/sim is the Quad Cortex from Neural DSP. I've seen Kiko from Megadeth uses it for his live shows.

I think the AxeFx has more virtual knobs, but the QC is easier to use, and has Kemper-like sound capture tech.

The QC is my go to pedal these days. I have a Helix (rack) and a Kemper (rack). While there have been some hiccups on software it really is a solid little box.

Yeah, I switched from HX Stomp to the first batch of QC. Very happy with the result.

AxeFX is absolutely insane in amount of control it gives you over your tone. I would say far too much control. Probably the best tool for a tone tinkerer.

Don't go leaving it exposed if you connect it up to the venue's WiFi, if you're playing live. ;)

Don't run wifi and real-time software at the same time.

What wireless technology would you recommend instead then?

none, of course

Lol ok. Meanwhile the rest of the industry is moving on. I mean, how do you even do an wired connection to a drone? A lot of times guitarists will perform on a stage and the clients wants aerial views without the cost of a crane. Do you just tell them you can't do it and have them go with a different company?

With special nonstandard proprietary wireless protocols is the answer you're looking for. The same shit wireless stage mics use. Or control over cellular data. I can guarantee you they aren't pairing them to a 802.11 router. Lol ok.

I'm a little confused by this thread - it seems like you suggest that the drone should be controlled by the SBC in an effects pedal, that the guitarist is using to provide low-latency audio processing? As a techie and perpetual-intermediate musician, that seems a bit odd...

But, incidentally, at my day job I'm working on an embedded Linux system where audio latency matters, and which may well wind up with a WiFi radio (where latency probably doesn't matter so much). So, I'd like to understand issues in this space.

The core audio driver does a pretty good job with this.

The only signals going over WiFi here would be controls for the virtual pedals.

Just the beacons you get from other routers around can be enough to create disturbances

Using wifi to trigger real-time software is often a bad idea, but using it to configure responses to future triggers isn’t necessarily so bad.

The Raspberry Pi 4 has separate USB hubs: one for WiFi and ethernet (an internal hub), and another for external USB ports. The USB port service loop will run with higher priority, so there doesn't seem to be any serious adverse affect.

The same isn't true for SD card access, which does cause dropouts. I've seen a video that suggests that disabling power management for the SD card hardware will correct the problem -- specifically that changing power state causes a 3ms delay. But I'm not quite sure how to go about disabling that on a Raspberry Pi OS.

. WiFi doesn't seem to affect audio latency. That's not true for Raspberry Pi 3, where WiFi and USB ports do run on the same USB hub.

Audience participation show.

I had sketched an idea for a web ui that would talk to a VCV instance, outputting signals to a real eurorack device with an expert sleepers module… need to keep hacking on that.

DefCon Big Band

Somebody should make a device that plugs into a fractal axe FX and hijacks the USB connection and exposes a web interface. Now that'd be cool!

There are a bunch of USB over IP boxes you can buy, so it depends on the m what you're looking for, port-wise. That plus a wifi router gets you what you're looking for.

Eg https://www.digi.com/products/networking/infrastructure-mana...

While not a guitar effects box another project that does low latency audio things with an RPI at it’s core: https://monome.org/docs/norns/

Open source version: https://monome.org/docs/norns/shield/

This is a neat little box, but pre-soldered ones seem to be only available on the used market. There are bare PCBs out there, but I’m not very confident with SMD parts.

This is the thing that bums me out with DIY audio: people come up with extraordinary designs, do a limited run and then never (or rarely) make any more.

Norns is made and supported by Monome, and it is very much supposed to be a Product That You Can Buy... except unfortunately it's based on the Compute Module 3, which has been unavailable for a good while. They have been available in small batches occasionally over the last year, and hopefully will be more available soon.

Surface mount soldering is not too hard. I can't view the BoM on mobile, but from the photos the soldering looks achievable for someone with experience soldering through hole. Take a look at the document here https://github.com/monome/norns-shield/tree/main/bom - if it's mostly 0805 sized components you should be fine. Even a few 0603 would be okay if you have good vision and a steady hand.

> people come up with extraordinary designs, do a limited run and then never (or rarely) make any more

I never bought a milkymist. There are no more being produced. The design has never been updated for modern formats, such as hdmi. I have no clue how to design hardware.

A big part of the problem is FCC certification.

If you ship parts, you can avoid it. If you ship it assembled, you'll need to spend tens of thousands of dollars getting it FCC certified.

Also Pisound (https://blokas.io/pisound/) which has the benefit of built-in din MIDI, but without the active community sharing software (you kind of have to build everything yourself with PD or SuperCollider). Some people have gotten Norns running on pisound, but I could never get it to work.

I do not know about Noms, but I have a Pisound

It has audio in so needs no external usb sound hardware.

It also has a programmable button. A simple idea but very useful

Sweet, was just looking for something like this the other day. My use-case is not for guitar, but just to offload some fx processing out of a DAW - a friend and I have gotten into jamming recently, and while you can do just about any musical production task these days with a budget laptop and enough patience, where you quickly run into limitations is running multiple realtime effects. I'm not yet serious enough about it to start spending $ on effects pedals (which typically cost hundred of $ each), or even to have a clear enough idea of what pedals I would want, but I know enough about electronics to realize most modern ones are just a glorified arduino with a a 500% markup, so a budget-friendly programmable swiss ay knife pedal would be a dream.

For my use-case the touchscreen is entirely unnecessary (programming it via a WebUI sounds more convenient anyway if you don't need to use it sans PC), which is inflating t he BOM by about 500%, and of course RPi4 is a uniquely poor choice of target platform at this particular moment in time, so seeing if it can run on a headless Pi Zero is definitely going on my endless to-do list. ;)

I use an ESP32-A1S board for that. I use this project https://github.com/pschatzmann/arduino-audio-tools/ and the platformIO vscode plugins.

Have a look at this example to see how easy using that library is : https://github.com/pschatzmann/arduino-audio-tools/blob/main...

I was thinking about something similar but wondering if you could do it over Ethernet and the VST API with low enough latency to be useful for a DAW. Or if you can figure out a way to make latency less of an issue all together and enable remote VSTs.

The 500% markup should also get you reliability and a decent chance that your tech is familiar with it were you ever to be on stage. Look around, everyone uses the same stuff, for the most part.

As an amateur musician, three things stand out:

1. I'd have a hard time seeing that small screen onstage, and my big foot would likely mash the wrong effects button. Others might find it easier.

2. There are tons of good, cheap effects boxes out there, and easy to find used. I like Pi boxes, but this seems like a homebrew replication of what's on the market.

3. All good boxes are low-latency, in my experience. It's a fundamental thing I think most players need.

What are they? Am I missing a joke here?

Regular guitar pedals are cheap and abundant.

Comment was edited after my comment to include the things that stand out, which were previously missing.

I agree with both of these observations.

> I'd have a hard time seeing that small screen onstage,

The project is using a 7" (17.7cm) LCD display. The Line6 HX Stomp is very popular and has a 2.4" (6cm) LCD display.

You’re definitely right. I think the draw of all of this is to make it yourself. The same could be said of people who make their own diy home weather stations or web servers. You could always outsource for the same thing that’s better, more frequently updated and probably cheaper when you factor in the time it takes to make. It’s just neat to make the tools that you ordinarily have to dish out catch to get. :)

As others have mentioned, I think the interesting thing here would be understanding the latency for processing the signal. Anything in the single digit milliseconds would be fantastic! I know at one point I was looking into Raspberry Pi and ended up on Pedal Pi[0], though I couldn't get the parts I needed to make it work.

I ended up using Teensy[1] and related audio shields[2] to get things working from a sound/acceptable delay perspective. But being able to get things going on a Pi would probably make more of the advanced input controls much simpler to implement simply from a OS support perspective (like in this project with the WebUI). The UI I'm seeing in this project looks great and it would be cool to potentially see something like kits/preinstalled images roll out for this!

[0] - https://www.electrosmash.com/pedal-pi [1] - https://www.pjrc.com/teensy/ [2] - http://blackaddr.com/products/

Also, the teensy audio library is great!

https://www.pjrc.com/teensy/td_libs_Audio.html

Stable USB audio with 3.9ms latency is possible (measured using a loopback cable). Not sure if this product manages to achieve that or not.

https://rerdavies.github.io/pipedal/AudioLatency.html

> A Low Latency Guitar Effects Processor

What's the latency? I can't find numbers anywhere.

Exactly

What is the expected latency of those cheap usb devices?

I had one that was unusable live. Noticeable latency. Have they gotten better?

IME it MUST be under 10ms

IME, if you can't find an important quantity like that front&center in an advertisement, then its value is going to be terrible.

Applies to many things related to an advertised product. Things like price, quantity, material, country of origin, standards met, certifications, scores, etc.

I'm still patiently waiting for future digital mixing consoles to do all processing in software on inexpensive x86 or ARM processors. Currently due to latency and reliability requirements all DSP work is done on dedicated chips or FPGA which brings up the BOM and engineering cost. They often have a small ARM/Linux module which is used for the displays and network control.

The CPU tech is here today, and modern general purpose processors do a good job of handling low-latency audio. Someone just needs to put all that together in a unified and stable package...

Modern CPUs are great if your watt (and thus thermal) profile is unlimited.

Not so great when they aren't.

Does this form a problem for digital mixing consoles? As far as I know, these already have fairly beefy fans for heat exhaust, so I don't know if they couldn't just add more airflow?

Not especially. I'm more thinking of things that have a potential of being near a live mic, where a cooling fan is a no go, at least without a defeat switch.

Modern consoles are plugged into the wall and are large enough for active cooling. For a 1U rack unit they can be cooled with small server fans; noise is not an issue as the amp fans are just as loud and the music will drown them all out anyways.

Most processing of audio isn't CPU performance limited... For most realtime audio mixing, it doesn't matter if you use 1 watt or 10 watts for your CPU - the big speakers will easily be drawing far more, and the performers time will be costing far more than the electric bill anyway.

> Most processing of audio isn't CPU performance limited

Processing, mostly true (though there are still some reverbs that can chew an awful lot of cycles).

Synthesis, however, is a different story. That is definitely CPU performance limited once you get into substantial numbers of voices/tracks/synths.

It's CPU limited in the sense that it's extraordinarily latency/jitter sensitive.

Not exactly what you're describing, but I've been running my band directly into Logic Pro on an M1 (both for recording, and live shows). Dry signals go through amp sims and effects processors, and then route to both a FOH mix and an in-ear mix all on the laptop.

Wish I had seen the OP's project months ago, but one bonus of the setup I describe is the ability to swap effects after the fact (by virtue of having the dry signal) and the ability to automate effects (so I can engage distortion etc as soon as we hit measure XYZ instead of having to click a pedal)

I've done live mixing with a laptop, DAW, and interface in the past and it does work but it's not something I would be comfortable with for an important show. Even with something like SAC (which is specifically designed for the task) the chance of hangs, crashes, etc. go up as at the end of the day it's just a program running on top of your OS. The setup and config also gets a bit hacky and you'll be the only one able to use it. As far as I know the only specialized system that does this is the Waves LV1 which has a dedicated OS running on top of x86 hardware for processing. While I haven't tried it apparently it works quite well.

However I was more thinking of mixers like the QSC Touchmix/X32/etc. where the DSP probably eats up quite a bit of the unit cost, and how the price could be significantly brought down if the innards merely contained analog I/O and converters all tying into a powerful SoC.

It's interesting you mentioned QSC. They're one of the few vendors in the installed audio space that have made the jump to COTS hardware: https://www.qsc.com/resource-files/productresources/dn/dsp_c....

Modern CPUs are surprisingly good at low-latency video as well. SIMD on something like a Zen4 core is a really big deal if used properly.

I've got some prototypes in C# that can draw a 1080p bitmap and encode to JPEG in under 10ms. Using single threading, socket mux servers and aggressive multimedia timers means my network delay is usually right at 1ms.

I feel like if you are just worried about audio, there is definitely enough bandwidth here to do what you need to per unit time.

While video needs more CPU power, it can tolerate higher latency than sound.

Fair point. Audio also has a tendency to require more serialized throughput in complex signal chains. Video is more trivial to chunk out and process in parallel.

Spot on. For context: 60fps gives you ~17ms of wiggle room before you start dropping/delaying frames. With 96kHz audio you have 0.01ms between samples. Drift above 0.05ms and you'll start introducing time domain issues in the human hearing range.

In other words, 'realtime' audio processing needs to happen 1700x faster that 'realtime' image work. Bandwidth isn't limiting factor, deterministic and uniform latency is that challenge for any signal as the sample rate goes up.

Drift really isn't a problem with audio cards. Latency is. To put things in perspective, it takes sound 1ms to travel 1 foot.

The general consensus is that guitar effects have to have no more than 10ms latency.

Poor word choice on my behalf. Drift was meant in terms of sync between different channels or parallel processing paths and in reference to GP. Guitar effects (generally single channel, sequentially processed) bypass that.

That 10ms benchmark is a good one though. At that time window you've reached a full wavelength at 100Hz and it's right about the point where well practiced humans (e.g. musicians) will begin to perceive delay. It's a fascinating intersection between physics/engineering and psychology as signal latencies make the jump from being perceived as timbre to delay.

> I'm still patiently waiting for future digital mixing consoles to do all processing in software on inexpensive x86 or ARM processors.

Harrison Consoles have done this for more than 10 years.

I cannot confirm this in the same way, but I think it also likely that both Lawo and Studer digital consoles do this, and also possibly Allen & Heath. All 4 companies run Linux internally on their consoles.

A better alternative to the Raspberry Pi which is more suited for musical applications (and currently much cheaper) is the Daisy Seed by Electro-Smith[0]. You can program it in C++ / Pure Data / Arduino and Max/MSP Gen~. The community is very helpful and there are plenty of examples to start with. They also provide a few options to get started with some knobs/buttons. I'm not affiliated with them, just admire the whole ecosystem.

[0] - https://www.electro-smith.com/daisy/daisy

But Pi4 gives you about 400% better CPU performance. And you'll need it for a serious effect chain.

I've had some amazing ( horrible ) adventures in low latency music stuff lately. It has made me think about going back to the hardware side of music production. Previously I was an ableton-only dude.

All of the vst plugins are CPU bound and even though i have a top of the line i7 and 32 gigs of ram, my computer slows to a crawl when editing even moderate sized songs.

Specifically, there is an nvidia bug that introduces latency to real time audio, making guitar and other live performance unplayable.

It really sucks! At least it has finally been ack'd (Increase in DPC latency observed in Latencymon [3952556]): https://us.download.nvidia.com/Windows/531.18/531.18-win11-w...

This has been a problem for YEARS. Hopefully they will finally fix it.

I don't know how this compares to your i7, but I dabble around using Pro Tools w/ about 15 or 20 tracks at a time with several effects running in unison, on an M1 Pro processor w/ only 16 gigs RAM, and I typically stay under 20% usage according to Activity Monitor.

I mostly play guitar and don't notice the latency in most effect chains that I use.

Yeah it should perform way better than it does currently. There is a driver bug for Nvidia cards that causes it to run slowly and poorly.

It is extremely frustrating.

Well done!

For those interested: a predecessor called the "Jesusonic" was once made by Justin Frankel (of Winamp and REAPER fame): https://www.cockos.com/jsfx/ https://wiki.cockos.com/wiki/index.php/Jesusonic_Documentati...

Very cool, I've ordered a Raspberry Pi touchscreen this weekend and it should arrive today. I want to make a MIDI sequencer with it, or at least play with the idea. I hope my old Raspberry Pi can work with MIDI (over USB) without too much latency..

I've seen it done with the Pico, where it acts as a midi device when plugged in to a host machine; not so sure about the main boards.

The pi zero hardware supports gadget mode and from there it's a bit more work to get it to enumerate as a midi device. The pi 4 supports this via the usb-c port.

https://raspberrypi.stackexchange.com/questions/71613/how-to...

Depends on how you want to utilize it. If you want to connect it to a computer as a MIDI device, you'll have to use an Rpi4 or one of the Pi Zero's. The Pi's < 4 can't go into "gadget mode". I bought a new v4 for this exact reason.

The RtMidi library is probably your best bet for getting started. I found the ALSA library to suit my workflow a bit better, but the setup is pretty obtuse. RtMidi is much more user friendly in that regard.

Also, look into implementing the Ableton Link library. It runs over your network, and is honestly astounding how well it sync's devices.

Can you readily get an rpi4 anywhere?

The Odroid C4 you can actually get is an alternative with 4 USB ports and an OTG that is gadget capable. Unlike older models, mainline Linux support is decent.

Ideally I'd sync it with Ableton's clock. I'm using Ableton only for tempo and multitrack recording. Maybe this is achievable without the RPi being a MIDI device? Sending messages over the network, like GuitarEffects which uses WebSockets. My RPi 2 doesn't support gadget mode (I should probably start looking for a RPi 4).

I'm currently using an Elektron Digitakt to sequence my analog gear - it's great - but unfortunately the DT is limited to 8 MIDI tracks with only 1 LFO per MIDI track to automate MIDI CC data. I'd love a Cirklon Sequentix but the waitlist is just too long (3 years atm, arghh).

I wonder if I can hook up something like a Midiface 16x16 (https://miditech.de/en/portfolio/midiface-16x16/) to the RPi. The Midiface is Class Compliant so maybe the RPi can use it natively..(?) I'm a bit worried about performance though.

Thanks for your suggestions, I'm going to look into them!

I’d say you can most definitely do that. You can get usb midi conversion cables for next to nothing and send an out signal to a midi hub or Daisy chain it. Midi is a seriously slow protocol so as long as the actual audio processing is happening on other devices and you’re just sending midi to them, you’re definitely good to go.

Any samples or videos of this working? I would like to have a feeling of the latency and how the effects sound.

There's a two years old small video on reddit...

https://www.reddit.com/r/raspberry_pi/comments/p0jkb3/guitar...

This is really cool. I may have missed it (am on my phone and just skimmed the readme) but what effects are available and can you build your own?

Seems like there are already a couple dozen pedals defined, and you can define your own using the Q framework...

https://github.com/Quinny/GuitarEffects/tree/master/pedals

I've found the RPi4B to be somewhat awful for low-latency audio usage.

My particular use case is simply playing MP3s read from mmc through an MBox1 on USB.

No matter how much irqbalance, isolcpus, taskset magic, it never gets absolutely perfect. It gets better, but there's always spurious delays exhibited as occasional pops and clicks in the audio output.

I'm hopeful that [0] will improve the situation, but haven't had time to really dig into it let alone build a custom bleeding-edge mainline kernel - which I'm not even sure supports all the Pi4B hardware.

It's asinine that an otherwise idle 4Cx1.8Ghz machine can't even play MP3s on a USB MBox1 flawlessly with zero special effort...

[0] https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/lin...

Reminds me of NeuralPi, which uses ML models to emulate real AMPs

https://github.com/GuitarML/NeuralPi

I gave been using NeuralPi not on a Rpi and I have found the results sometimes to be on par with neutral DSP.

I've been running NeuralPi on a Pi 4. Fabulous!

https://mod.audio/ (I'm not affiliated - just a long time follower)

This should really be at the top of the thread, since it is fundamentally a commercial version of (and which also substantively predates) the project in TFA. Note that most of the technology in a mod is FLOSS (if not all of it).

This looks very promising. I'm interested in how to make effects and learn more about it. What a cool project!

From my understanding the Line 6 Helix uses two 450MHz SHARC processors, ADSP-21469. Other effect/amp modelers use more or newer of the same family[0].

Can anyone comment about the relative processing power of a RPi vs the market solutions? Is the RPi theoretically good enough that a pedalboard could be completely modelled?

[0] Interestingly, it sounds like SHARC chips were designed to be low cost processors for single use applications in guided artillery shells.

According to Analog's product page [1] for the ADSP-21469, it delivers 2.7 GFLOPS at 450 MHz.

And Wikipedia's page of all the RPi models says that the latest (4B) manages 8 GFLOPS at around 1.8 GHz.

If that means that the answer to your question is "yes" or "no" is unfortunately a lot harder to Google and/or figure out. I would assume that the SHARC-based devices run on the metal, whereas most applications for the Raspberry Pi run under Linux, for instance.

[1]: https://www.analog.com/en/products/adsp-21469.html#product-o...

[2]: https://en.wikipedia.org/wiki/Raspberry_Pi#Simplified_Model_...

Edit: grammarish.

I'd love to see one of these Raspberry Pi guitar FX projects run on bare metal or a RTOS instead of Linux.

There's also a real-time version of the Linux kernel which is meant for exactly this type of application.

Most of the real-time Linux kernel has been ported into mainline. Mainline is perfectly capable of producing stable low-latency audio, as long as you're running on threads with real-time priority.

The real-time kernel provides additional improvements. But it's incredibly difficult to find up-to-date real-time kernels these days.

Also worth mentioning that there are enormous USB audio improvements in kernel version 5.10.0, both for performance, stability, and compatibility. As far as I know, there are no real-time kernels available for Pi 4 with a kernel version greater than 5.10.0, and building one is painfully difficult.

Generally this type of work is done on DSPs and FPGAs, because they can generally get much lower latency than a CPU. While this can get latency low enough that nobody will notice, it is still there, and there isn't room for anything else in your signal chain to also have latency and still be unnoticed.

They will still have a CPU on devices like this, but all it does is run the UI, the sound processing is not done on a CPU.

Why would a DSP be able to get lower latency that a modern general purpose processor? DSPs might have been a thing 15 or 20 years or 30 years ago; but I cannot imagine anything that a DSP would do better than a modern ARM processor.

> The root of it is the Transform function which takes an input signal, performs any kind of transformation, and returns an output. All input and output values should remain in the range [-1, 1] otherwise you'll produce some really gnarly popping and cracking.

So you have to make all your own effects in code? It would be cool if it connected to something like Guitarix (open source) so you could use existing guitar effects. (Disclaimer: I've never used Guitarix, so it might sound shit)

Btw, for anyone who doesn't play guitar, but is interested... Gone are the days of those huge pedal boards and having to buy 30 different pedals. Emulation is getting really really good. You can either buy a multi-effects pedal and use that onstage. Or if you're in the studio, NeuralDSP is software which can emulate basically any sound you're after. It's expensive though, but it sounds better than free alternatives like Amplitube.

Nice, I just installed this on a Raspberry Pi 3 with a touch screen I had around. Save for an issue when installing rtaudio (requires libasound2-dev in Debian), it went almost flawlessly. I'm guessing the RPi 3 can't handle realtime audio and graphics, as popping was frequent (although not annoying for the purposes of experimenting). The buttons did nothing, I'm afraid, but the web interface is ace!

I'd love to write about this soon. Kudos to the coder

I thought is hard to have low latency audio in Linux with stuff like PulseAudio.

Possible with jackd etc

I switched to BSD and got much better latency on the same hardware

But I think that was the alsa -> BSD oss driver swap, which is possible on linux too. I had just never tried it, since I assumed linux was best

(I'm only sharing this cus i think its interesting, and may help you get good latency)

I'm enjoying the downvotes :)

Was anything I said wrong?

Seriously, I've spent a long time fixing up low latency audio

jackd + OSS on whatever OS u like is not bad advice fyi

I honestly have no idea.

Personally I'm very happy to see that info as I do use both Linux and *BSD and wouldn't have ever tried setting up BSD for audio until now.

BSD has in-kernel audio mixing, IIRC.

I've begun tinkering with GhostBSD lately just to learn more about BSD. Pleasantly surprised to find a lot of familiar linux audio tools were ported or can be easily made to work. Also happy that my ancient Presonus AudioboxUSB works fine in BSD. So much so that I get less random audio glitches than I do under windows 10.

The RTAudio tools this is based on allows different backends to be configured. I guess you could use Alsa or Jack if you want...

https://github.com/thestk/rtaudio/blob/master/install.txt

Jack offers the low latency necessary for music production.

Pipewire is supposed to offer low latency as well - at least lower than Pulseaudio, but I don't know how much.

Pipewire runs neck-and-neck with Jack and Alsa. The benchmark pretty regularly; and the difference is +/- nothing. Pulseaudio isn't appropriate for this application.

Nah, low latency audio is absolutely possible, but not necessarily using pulse audio. Using either Jack for the low latency stuff has been the advice for years, but with the advent of pipewire you can mix and match without too much difficulty.

A quick skim didn't turn up what audio backend this project is using, but I'm using patchbox with modep on my Pi4 as a bass pedal board and it's pretty much flawless low latency. I do need to add a fan to my pi since it's mounted underneath a pedal board and doesn't get quite enough air though.

Sep 2 of the build instructions says it requires RtAudio.

RtAudio is just an API wrapper which calls JACK, Alsa, etc behind the scenes and does not add latency, just the overhead of a couple c++ method calls

what is the latency of a raspberry pi running this type of software? can't find figures anywhere

3.9ms measured audio latency using a loopback cable.

https://rerdavies.github.io/pipedal/AudioLatency.html

I also wonder about the latency of the behringer interface. I've had one before, the sound quality is pretty good for entry level high-fi audio, but when I've run the audio into a DAW and back out post-processing there has always been an audible delay.

Granted, this was in windows, but from what I understand there is always going to be some audible processing delay with a USB 2.0 interface.

Wow, what fx are you running -- and does that or could that include convoluted reverb?

(Convoluted reverb is utterly awesome but more processor intensive than almost anything else)

Edit: and sorry, to be helpful, by low latency, could you meantion how many ms that is -- because while I'm very happy for you, and really interested, 15ms is very different to say 5ms or lower

My normal chain is a noise gate, followed by a comp, a little emulated tube drive, an SVT amp model, an impulse response modeled cab simulator (of course running a 8x12 SVT cab), a plate reverb, and a limiter. I've got some crazier ones, but as a beginner bassist, that's all I really need. There are literally hundreds of effects to choose from though.

Convolution reverb is an option, but it ended up being a little more than the Pi4 could handle iirc. I didn't tinker with it much though.

Latency is under 10ms, I'd have to go back and check my settings to confirm exactly though.

Just double checked my jack settings; I'm running my interface at 48000hz with a buffer size of 128 and 2 periods for a theoretical latency of 5.33ms. I'm sure USB adds a fraction of a millisecond as well.

Thanks! Musically that's really viable and fun setup

A really helpful data point :)

You're doing great! You don't need my help, but just share my own path, because it's fun :)

Yeah, play the bass, if i start talking about computing im not helping...

Err, why am i still here?

I switched to BSD or the OSS driver and went from massive effort ~5ms to zero effort ~2ms

Then I made a two button controller and pedal->usb, and started programming an arbitrary effects controller: a looper + buttons to switch the pedal to control any fx parameter

Fun, but it didn't help me make better music :)

You'll get lower latency with 3 buffers. Try 16/3.

PulseAudio is an audio server aimed at desktop applications. It has nothing to do (or offer) low latency audio work, and would never be used for it. ALSA is the driver layer on Linux, and it can go as low or lower in latency terms as the audio driver layer on any other platform.

There is also Pisound for low-latency guitar effect processor which is a hat for the Raspberry Pi. https://blokas.io/pisound/

The UI, MODEP, is based on the Mod Devices work on their open source pedal (Originally Mod Duo) > https://mod.audio/.

I bought a PiSound to tinker with the Midi interface a bit, had no idea it was this powerful.

This is a wonderful project, thank you for sharing your process! I'm excited to try making a cute/custom FX brain with this.

This is awesome. I'd be really interested to find out if a Raspberry Pi can run an open source amp modeler like NAM (https://github.com/sdatkinson/neural-amp-modeler).

I'm not sure if the GPU is powerful enough on the RPi, maybe something like a Jetson Nano could handle running ML models better.

Very nice! A screenshot or two of the web UI would be a good addition to the readme

Also see: https://rerdavies.github.io/pipedal/

which solves the problem with using a bar's Wi-Fi by using Wi-Fi Direct.

What is he using for actual audio interface? RPi has something awful onboard and none of the pictures show anything external. USB would add 0.25ms latency just for being USB, best case.

Round trip latency for Raspbery Pi 4 USB audio measured using a guitar gable from output to input: https://rerdavies.github.io/pipedal/AudioLatency.html

Stable audio at 3.9ms latency (measured using a loopback cable).

If you are interested in this see https://blokas.io/modep/

Hosting plug-ins is a very powerful way to go.

Looks a lot like https://mod.audio/dwarf/ which I own and love.

Nice project, but nonway on earth I am taking anything with a breadboard and jumperwires onto stage — except maybe if I plan to make it part of the performance

I'm guessing the harder part would be actually finding a Pi for this?

have to compare with Bela when it comes to latency

Surprised that a (local) Web call can be considered low latency. Isn’t this why people use grpc etc?

I'm assuming that the Web UI is just for changing settings.

The low latency would refer to the latency of the audio input (guitar) being processed and producing a sound. Ideally you want no discernible delay between when you hit a string and when a sound is produced.

As for how much delay is considered "acceptable", I'm not going to open that can of worms...

Reply on news.ycombinator.com