THE VALUE PROPOSITION IN AUDIO: ROLLING YOUR OWN IS AS EASY AS PI
Recording and working with audio files on a simple Raspberry Pi DAW
Yes, a Raspberry Pi can record music well. My album (right) was the original project, but Covid closed the club last Feb. So I made this FLAC (download) master in Audacity after normalization & mastering from 6 stereo 32/44 wav tracks recorded in my home studio on Ardour with a $100 M-Audio DAI. The same Pi 4 did everything, even monitoring for near zero latency & no need for correction of any kind. I’m playing all the instruments.
The acoustic guitar was mic’ed with a stereo pair. Keys, bass, Roland synth vibes, electric guitar, and synth drums were recorded in stereo directly through the DAI.
Studer, Mara, JoeCo et al aren’t worried about being replaced by Raspberry Pi. But you can make wonderful live recordings with one and have great fun doing so!
Music is the common thread among audiophiles. There are at least as many subcultures among us as there are foci of interest and enjoyment. Collectively, we enjoy (but are not limited to) studying, designing, building, playing with, admiring, experimenting with, talking about, debating, and/or otherwise enjoying music and the myriad of equipment with which to capture, store, and reproduce it.
Many audiophiles are quite satisfied sitting before their speakers or under their headphones listening to their favorite music. More than a few give absolutely no thought to the processes of recording, optimizing, and presenting it – they do not care what went into it because their focus is solely on the endpoint. But at least as many dive into the deeper waters, motivated by a range of goals from better sound quality to fuller understanding of why things sound as they do.
Kant said that “[e]xperience without theory is blind, but theory without experience is more intellectual play.” Although not a disciple of Kant’s (or any other) categorical imperative, I think he was onto something important for audiophiles in this particular quote. AS has an objective-fi forum for deeper exploration of both the dichotomy separating subjective experience from objective theory and the many ways in which we attempt to relate the two. Further, the entire site is peppered with posts about defining, achieving, measuring, hearing, and appreciating “accuracy” in recorded music. And we have more than a few participants / members who are highly experienced and quite competent in one or more of the many disciplines that affect accuracy, realism, SQ, and/or whatever other euphemisms you prefer to describe how well reproduced music reproduces the source material in your listening environment.
But many audiophiles have opinions borne of a combination of their interpretation of technical data (which they may or may not understand and which may or may not be useful), the published or otherwise expressed opinions of sources deemed to be expert in accord with a variety of differing and sometimes inconsistent parameters, casual anecdotal reporting, etc. They lack hands-on interaction and any confirmation that their interpretations are correct. Deciding that you are hearing the effect of an anomaly like jitter or a specific perversion of the sound of a recorded instrument is difficult to do correctly when you have no idea how the phenomenon in question affects (or does not affect) what you’re hearing.
Trying to assess the true, literal accuracy of a recorded performance without knowing how it sounded when performed is like trying to fix a hernia after studying the procedure in a surgical textbook or trying to identify the kind of oak barrel in which a wine was aged by reading about it. Some knowledge is simply not empirical – you have to acquire it through education, effort, and the resultant experience.
Thanks to the advent of inexpensive but powerful small computers and digital audio interfaces, audiophiles can now experience first hand many audio events that were until recently inaccessible without costly and sophisticated equipment in a dedicated setting. You can make excellent live recordings easily with less equipment than would fill a cigar box. You can rip and archive your own CDs and vinyl with it, and you can manipulate audio files in many ways to both learn how changes sound and find the ones you prefer. You can also analyze those files to gain a better understanding of the relationship between measuring and hearing.
The knowledge and experience gained by working directly with audio files is invaluable – and it expands the audiophile’s ability to hear, interpret, understand, optimize, and thus enjoy his or her music and equipment. This article is aimed at those who want to learn more about working with sound files by making manipulating them for fun, knowledge, and pure pleasure. Let’s get started!
OUR PROJECT TOOL KIT
We’ll be assembling a few tools into a versatile minilab with which you can record live audio both at home and in the field, capture and convert prerecorded files (e.g. rip vinyl to wav, convert wav to FLAC, convert video sound tracks to audio files, etc), and manipulate audio files in various ways. In the next installment, we’ll get to using a Pi for audio testing, display, analysis, and other technical purposes like DSP None of the assembly required to turn a Pi 4 into a digital audio workstation (DAW) is sophisticated or difficult for anyone who can manage software installations and updates, get music from boxes connected by combinations of USB, TOSLINK, RCA and HDMI cables at various points, and use even the most basic features of Volumio, Moode, Roon, JRiver Media Center, VLC etc.
You don’t have to build anything to learn from this article – but doing facilitates learning, and you’ll end up a more knowledgeable audiophile with the experience you’ll gain. And you’ll have an inexpensive device that you’ll use for one or more audio things you’ve always wanted to do but haven’t gotten to yet. I use my project Pi for live stereo recording, editing audio files from others, ripping videos to FLACs, ripping my vinyl, and playing backing tracks with which I practice my instruments. I’ve also built it into a Pi-powered guitar effects pedal, an active crossover, a free standing WiFi hub, and a spectrum analyzer (the last in progress using GNU Radio, which may or may not work for audio frequencies – we’ll find out!). But first and foremost is the creation of the basic DAW and a simple ministudio setting in which to use it.
You can do all of this with most of the good SBCs currently on the market, as long as their specs are comparable to a Raspberry Pi 4 with 4 gigs of RAM and you can run a fairly robust Linux-derived distro. If the performance specs meet minimum requirements for the DAW software of your choice, it’s a go. I did all this with the latest 64 bit Raspberry Pi OS, which is still in beta but is much more complete and reliable as I write this than it was when I wrote the Hot Rod Pi article a few months ago. It’s not yet on the main Raspberry Pi Foundation download page – you have to get it from this link (which is an official RPF source – never fear). You can do most of the things in this article with the current 32 bit Raspberry Pi OS, but you have to have the 64 bit OS to access the full 4 gigs of RAM. I also haven’t tried an 8 gig Pi 4 yet because 4 gigs has been plenty and $65 for a Pi is pushing the value envelope a bit if it doesn’t add to utility and/or quality.
I’m using a 250G USB SSD instead of an SD card, and ZRAM is enabled. So the extra 4 gigs of RAM is probably not worth the cost for these uses. With a cooled case, a simple SSD (like the Inland house brand at Microcenter, which has worked very well for me) and a PS, you’re pushing $100 even when they have a sale. You’ll need a microSD card to set up the Pi so it will boot from USB, which is easy. And you can run directly from the SD card if you don’t want to bother with a USB SSD. The disadvantages are few - it’s slower, has much less capacity for the file system, and will wear out over time with continued use.
If you want to use an SD card for the boot and basic file system, make it a good one like a Samsung Evo+. Any card with a V30 rating should be quite fine for audio use. I’ve gone as high as 128G just to see if it would work, but a 32 will do the job fine as long as you’re using network storage or a USB drive to park your files. You do not have to boot from the USB drive to use it for storage. You really don’t want to be accessing an SD card for audio operations, since the life of the card is proportional to the number of data accessions (among other things. If you do use a card for everything, make sure you back it up. Cloning it to at least one duplicate is a lot easier than using a backup scheme, and you can just swap the card to be back where you started minus whatever data you added or changed between the last backup and failure.
RUNNING ALTERNATIVES TO RASPBERRY Pi OS
You can run many Linux distros on a Pi. There’s even a full Ubuntu 20.04 LTS server package for ARM processors that runs fine on a 4 gig Pi 4. I’ve tried several and found them to be surprisingly capable when run from a USB SSD, even with only 1 gig of RAM (as long as you have ZRAM active – see my AS article on hotrodding the Pi for more info on tweaking your Pi for better performance). The simple truth is that I can identify no sonic differences at all among identically spec’ed FLACs and wav files created on a good Pi regardless of the software and OS used. So bigger is definitely not better when it comes to software packages on a Raspberry Pi, and I see no reason not to use the 64 bit Raspberry Pi OS. You can do whatever you want to do (within very wide limits) using Ardour, Audacity, and/or one of the other DAW packages I’ll bring into the discussion in a few paragraphs.
There are several DAW packages available from one or more of the Linux repositories, so you can install with the package management program in whatever flavor Linux you’ve chosen. Ardour and Audacity are available from every distro-based repository with which I’m familiar (which is a lot of them). Unless you choose an oddball distro, you shouldn’t have to add additional PPAs.
Once you have the OS image burned to a microSD card, you can prepare, configure, and operate a Pi without another computer. Just connect it to a monitor via HDMI, and plug in a mouse and keyboard. OS installation and setup from the SD card are very easy, as is cloning the configured card to a USB SSD with the Raspberry Pi copy utility already embedded in the OS. The alternative is to SSH into the Pi with another computer on your network and/or set up the remote access software of your choice (I prefer VNC) . You can do everything after initial boot from another computer or device, one way or another. Controlling your Pi DAW is easy from any device on your network with a VNC client.
As a value driven guy, I love the true bare bones approach. You can even set up your Pi using SSH from your mobile device, as long as both devices are on a WLAN. With an SSH app like Termius (there are several available from the app stores, but Termius is the best I’ve found for mobiles), it’s easy as Pi. And I (like almost everybody else) prefer Putty for SSH from any computer.
INTRODUCTION TO AUDIO INTERFACES
Now that you have a Pi ready to go, you need to decide how you’ll route audio into and out of it.
The Pi has I/O audio interfaces that you can use out of the box, although you probably won’t want to do so unless you’re a podcaster with a decent USB mic (more about microphones later). There’s no analog input, but the USB bus is quite fine for audio use (both I & O). You can use USB microphones directly,and they work surprisingly well. But limitations on USB mic input start with its “directness”. Less expensive ones have no preamp, buffer stage, or other opportunity for signal control or manipulation. So the mic sounds like the mic sounds, and there’s nothing you can do about it until the signal reaches the DAW. This is not a terrible thing, and all of the DAWs in my little table have good DSP and other plug-in utilities to do this. You can make a virtual MC device from multiple USB audio interfaces by editing your ALSA configuration file. Even better, you can plug 2 or more USB mics directly into a Pi and they’re recognized independently in a DAW with MC input capability. All USB mics have onboard gain stages and ADC, but only the best ones include controls on the mic.
For analog output, there’s a rudimentary DAC within the Pi’s SoC that drives a 3.5mm stereo jack. The SQ’s not great (to be polite), but it’s better than many mobile devices with marginal audio and works for monitoring rips and general recordings requiring neither constant attention nor critical SQ.
Although you can record live audio directly with USB mics, few serious audiophiles will do much on a Pi without “real” mics and a true digital audio interface (DAI), which is a combined ADC/DAC. All but the least expensive DAIs have built in mic preamps, many with integral phantom power necessary for powering condenser and reference mics like my Dayton EMM6. You can use HATs too. HIFiBerry makes a DAC/ADC that’s quite competent, great value, and efficient in every way (but without a preamp). There are many Pi HATs available for audio I/O – Google them if you don’t like my choices.
There are many simple, decent ADC/DAC interfaces on the market for musicians. The big music retailers all have them on their websites, as do major online retailers like Amazon. Even a $35 plastic USB dongle like the Behringer UCA222 (below left) is surprisingly good. The dongles are OK for live audio and much better than OK for ripping vinyl to Redbook. Units like the U-Phoria (below right) are the next step up. For about $50, you get a very nice, fairly quiet mic preamp and phantom power.
You can get a seriously fine, entry level“prosumer” quality 24/96 USB ADC/DAC for as little as $100, like the PreSonus AudioBox 2x2 ($100) and get 24/192 specs for only a few dollars more in units like the M-Audio 2x2 that I use, which has a very nice balanced mic preamp w/phantom power and great SQ going in and out. You can even go up to 32/192 with great SQ plus fine mic preamps w/phantom power and stay under $200 with excellent units like the Steinberg UR22c ($189) pictured below.
So there’s a world of high quality audio interfaces at true value prices for the audiophile who wants to be able to do more than just listen to audio files. Even better, the DACs in these are good enough for many of us to use for general listening with little or no sacrifice compared to some much more expensive units marketed for audiophile use. Spend $300+ on a DAI and there’s a fine little DAC in it.
If you’re going to use your Pi or other SBC for general listening as well as recording and other audio functions (which is perfectly fine and easy to do), you’re probably better off buying a digital audio interface than getting an entry level DAC, as long as you limit your audio adventures and enjoyment to 24/192. I don’t know of a DAI under $1k that will do DSD, although there are some fine pieces coming in at $1500+. I haven’t made any audio files on a Pi beyond 24/192, which is fine for stereo ripping and live recording. I also haven’t tried simultaneous multichannel recording on the Pi 4 because the 64 bit OS wasn’t up to full feature audio until recently. MC recording works fine on the 3B+ with the Octo card I discussed in the MC article in this series about 6 weeks ago, but I haven’t done it with a 4 yet because covid has shut down all the venues in which I’d record live bands. For the multichannel recordings I make playing multiple instruments, I lay down additional tracks one by one along with playback of the “scratch” tracks, which are usually just a metronomic click for timing and a rhythm section part (most often a bass line).
If you want to do multimic live recording, there are now some seriously wonderful DAIs with multichannel digital mixers for very little money. The Behringer XR18 is a $600 18 in / 18 out mixer with WiFi, Ethernet, and very low latency USB connectivity. Although a bit more full featured than most of us will ever need, it’s a perfect device for the audiophile who wants to record local concerts and home performances on a small, easily transported and set up device like a DAW on a Pi - and it’s truly great value. I’m sure there will be more and more digital mixers with integral DAIs and network audio capability coming on the market in our price range over the next few years.
If you really want to get into high res recording with a Pi or other SBC, you may be able to use one with a serious processor on a Ravenna network. Yes, this is overkill for a Pi – but I’ll bet it’s fun & I plan to try it. There is now open source software for the Pi that supports AES67, an audio-over-IP / audio-over-ethernet protocol. Here’s a list of currently available Ravenna-enabled audio products.
As much as I love using SBCs for everything I can, I strongly suggest considering a stand-alone recorder like the TASCAM 3000 if you really want to get into DSD recording. For less than a grand on the street, it’s very hard to beat this machine and others like it. So this article won’t discuss making or working with DSD any further. If we get enough interest, I’ll work up a summary of these devices.
THE MAIN TOOL IN OUR BOX: THE DIGITAL AUDIO WORKSTATION (DAW)
The definitions of a DAW vary across the internet. Some consider it to be software, others hardware, and others a standalone SW/HW package for all audio manipulation, including recording, mixing, mastering, format conversion, video integration etc. This widely used description works for me: “an electronic device or application software used for recording, editing and producing audio files”. And that’s what we’re going to make, so you can get your hands wet. In doing so, you’ll hopefully have a lot of fun, learn something useful to you, and spend very little money.
You can do most of the things we’ll explore together on an earlier and/or less capable version of the Pi (e.g. a 3B+ or a 4 with 1 or 2 gigs of RAM) or another SBC. But processing power in any of these is marginal even for 2 channel recording if you want to monitor through the same device (which has advantages we’ll discuss in a little while) and use any signal processing at all. Fortunately, recording to wavs yields excellent files that you can post-process to your heart’s content on the same Pi. My DAWPi is a 4 gig 4 overclocked to 2147 in a fan cooled case (mandatory to keep it running well for a long time). It is the minimum Pi that’s capable of simultaneous duplex recording & monitoring from the DAW software without sonic aberrations from pops to drops, and Ardour is the software of choice.
There are many famous DAW programs out there, like Cakewalk, Sonar, Ableton, Cubase etc. Sadly, the big name products in this arena are all for PC and Mac – Linux is a stepchild in this family, which is bizarre given the large and growing popularity of Linux for general audio use. You can run many of the Windows programs on a Linux box using WINE, but even a really hot Pi lacks the power to do this well and I don’t recommend it.
Look to the open source community for audio recording and manipulation software for SBCs. There are some incredibly powerful open source DAWs out there for Linux, and they offer 90+% of the functionality and usability of the big guys with few exceptions (none of which is a deal breaker for me). The big 3 are Ardour, Reaper, and Non (a novel approach that’s well worth exploring if you’re curious). Audacity is a recording program with versatile editing features, rather than a full tilt boogie DAW. It’s a stellar program that I highly recommend, but it lacks a few DAW features that you might want, most notably the ability to use virtual instrument plugins. LMMS is a fifth alternative that’s getting better rapidly for a Pi, but it lacks a few basics like VST plugin capability.
I believe the choice of software is clear for the two things most audiophiles will want to do with a Pi audio workstation. The only audio recording packages in the running for starters are Audacity, Ardour, Reaper, LMMS, and Non. VLC is not for recording, but it’s great for extracting audio from videos – I use it regularly. Use the table below to consider the options, of which there are few others from where I sit. You can do what I do and periodically Google “open source DAW for Linux” to see what’s new. If something catches your eye, download it and try it. I’m sure there are fine packages I’ve never seen and there will be more to come every year.
Audacity is my first choice for fast & easy live recording with simple stereo mic setups. It’s also my choice for editing, duplicating, modifying, or ripping audio (from vinyl, audio CDs, other disc formats, videos etc). Note that live recording of concerts, shows etc differs from studio recording in which multiple parts are recorded one at a time, for which I prefer Ardour. The strengths of Audacity are easy to enumerate: it’s proven itself over many, many years to be reliable, consistent, versatile, and easy to use (for a DAW – it does have a bit of a learning curve, but it’s soooo much easier to master than most of the others); you can fairly easily integrate multiple tracks and export as MC files, although you have to download and add the ffmpeg library (which is very easy to do). Audacity itself doesn’t support native MC formats because of patent issues. But you can assign specific tracks to individual channels and export as a standard MC scheme (eg 5.1) as well as one of your own creation.
Audacity’s weaknesses are also easy to enumerate: you can’t import video files so you can’t extract audio directly from them (which only VLC does easily); you can’t use real time DSP – you can only apply the extensive processing capabilities to recorded tracks; you can’t play VST instruments directly in Audacity. Although it does support VST plugin effects, Audacity doesn’t support VST plug-in instruments, so you can’t just add a USB keyboard and play MIDI musical instruments (sampled, synthesized, etc) directly in Audacity; You can direct the output of a stand-alone instrument or rhythm machine like the great Hydrogen drum / rhythm machine to tracks in Audacity, but you can’t integrate it directly with Audacity and use one set of controls for both like you can with Ardour; Audacity only has a pair of level meters – so you can’t visually monitor recording levels of more than 2 tracks; finally, loopback monitoring pushes the CPU too hard, while Ardour lets you monitor in real time on a 4.
Apart from the limitation of one stereo pair of level meters, Audacity is fine for live recording. It will lay down as many tracks as the DAI and drivers will enable, up to 16 at a time and a max of 48. I doubt that a Pi will process anywhere close to 48 tracks & haven’t come close. Eight is probably a practical limit for smooth, glitch-free recording, editing, and playback even without DSP or other processing add-ons. Using the 6 in – 8 out Octo sound card I wrung out thoroughly in a recent article in this series, I was able to record 8 simultaneous live tracks easily. But audio monitoring of the program via loopback through the Pi while using Audacity rapidly led to extraneous sounds that would please a Cage devotee. And when activating the dropout monitor (another very useful feature of this program when using a limited processor like the Pi’s), dropout alert windows popped up every time I tried to audibly monitor what was being recorded. As I’ll discuss further in a while, the only way around this is to carefully set your mixer levels during a thorough soundcheck and audibly monitor live recording from your mixer’s monitor bus while visually monitoring your level meters / indicators very carefully.
Audibly monitoring the input only lets you hear what’s going into the recording, not what’s been recorded. Those of you who remember tape recording can visualize the difference between a 2 head machine and a 3 head machine. Two head machines had no way of immediately playing back what was being recorded – so recording quality could not be monitored until the recording was complete (by which time it’ may be too late). The third head allowed playback in (almost) real time sync with recording – the slight delay was because the record and playback heads were physically separate.
This is a limitation that’s potentially serious only in that you won’t be able to detect a failure affecting recording quality in real time, a rare problem if you’re meticulous about setting up and maintaining your equipment. It’s not hard to watch your equipment and your level meters closely, except that Audacity only has a pair of level meters in the GUI – so you cannot monitor record levels on more than 2 channels. I use this approach when recording on a Pi because I have no choice. This is a limitation of SBCs, not of DAW software. And it’s mainly a problem for live recording – we’ll discuss ripping a bit later on, but the need for close monitoring is not as critical when ripping as it is when recording a live performance that you can’t just repeat if your recording is not good.
What could go wrong that you’ll miss by just watching the level meters? Many failures can affect SQ either by degrading the signal or adding extraneous noise without causing a noticeable change in strip or LED level meters. A single fluorescent light turned on in the middle of a performance can introduce buzzing undetectable without audio monitoring of the recorded signal. And an unknown “silent” process that hits the CPU can cause dropouts in the recording. So Audacity is not the best choice for live recording of one-off performances. Done with care as above, you can get away with it – I have.
Ardour is another very nice, fairly light DAW package that’s ideally suited for use on an SBC. The GUIs are significantly more sophisticated than Audacity’s, although a lot of what Ardour does can be done in Audacity with a little effort and flexibility on your part. But a lot of what Ardour does simply cannot be done at all in Audacity. Ardour is a better choice for live recording of critical programs because you can monitor the recorded signal directly through it on a good Pi. It’s also easier and better (IMO) for dubbing and laying down multiple tracks one by one. Its latency management is also a bit better and easier to use, although you can also correct for latency in Audacity.
Setting Ardour up for a session is not as simple as doing so with Audacity, because you have to select the input and output devices and busses for each track in your project. Accidentally connecting the input and output of the same bus or device will cause feedback (for which there’s thankfully a red indicator in the GUI). You can preset and save these parameters as a project, opening the same one each time and deleting the tracks in it after saving the session as a named Ardour project and exporting the audio files individually for backup and security.
Ardour can use VST plug-in instruments, which is a major feature for musicians. Unfortunately, VST plugins do suck up processing power and are therefore tough on SBCs. But laying down one track at a time with little or no DSP, a strong Pi will do OK. I can get a pretty fair Hammond organ sound, some fine keyboards, and a host of “lesser” instruments. Hyrdrogen (an excellent open source drum machine that’s installable from the Debian repository using apt) works very well with Ardour on a Pi.
The main limitation for virtual instruments on a Pi is that VST is a Windows platform. The availability of plug-in instruments for Linux is much less (although improving daily) and there are now a few VST “host” programs that will let some VST instruments work on Linux. There are other plug-in systems like LADSPA (Linux Audio Developers’ Simple Plug-in API), but the availability of instruments is still quite limited compared to the plethora available for VST. It can take a bit of work to find virtual instruments you like and make them work in a Linux DAW.
Ardour will synchronize with MIDI sequencers. You can control rhythm tracks directly from Ardour’s transport controls, and it supports external hardware control surfaces for mixers, virtual instruments, etc.. It supports more automation than Audacity and has a number of useful real-time features like changing plug-ins on the fly and moving samples to different tracks while they’re playing. Ardour is a more complete and comprehensive DAW than Audacity, so it can do more for you if you need more (which many audiophiles do not). Finally, Ardour has a built-in mixer GUI with a level meter for each and every track in your project plus great management of all plug-in instruments and signal processors.
All of this makes Ardour an excellent choice for live and studio recording with more than 2 mics. It’s OK for ripping vinyl, but Audacity makes the process much easier in many ways, e.g. with simple labeling of individual tracks and exporting of a tagged and titled album right to your library. Ardour’s other significant deficiency compared to Audacity is how difficult it makes simple editing tasks on your recorded files.
LMMS, Non, and Reaper are other interesting and useful DAW packages whose GUIs and feature sets you may prefer. But I’d urge you to check out Audacity and Ardour before looking further. They’re both gold standard packages that make an unbeatable pair for most audiophiles who want to do anything with audio files besides listen to them.
Most of you know VLC because it’s the default media player in many Linux distros including Raspberry Pi OS. It’s a very versatile media tool that does almost everything an audiophile could want, except that it does not record audio. Still, it’s a valuable tool in our kit and you should be able to live with it very nicely if you have no interest in making or manipulating your own audio files. One of its most endearing features to me is its ability to convert among many file formats. It even converts video files to FLAC or wav (Redbook), so I can enjoy music from concert videos along with the rest of my audio files once I edit and supplement the tags it sets when it creates the files (which you can control).
JACK is an audio connection manager. Think of it as a virtual patch bay that lets you connect ins and outs of software, hardware, and virtual devices in any configuration to make the signal chain of your choice. It can also be the master clock for your DAW, rhythm machine etc. It’s a bit complex, but it works great and is worth the effort to master. I strongly suggest you use it. Here’s the web page.
Here are the choices for file type and codec – you can select as you wish for conversion, or batch it.
PERIPHERALS FOR AUDIO RECORDING
MORE ABOUT AUDIO INTERFACES
The digital audio interface is the final common pathway into and out of your DAW, which makes it a critical piece of gear for any audio recording task. It doesn’t matter if you’re recording live performances, ripping vinyl, making an audio file from the soundtrack of a video, laying down multiple tracks yourself for a demo or a one person band, creating a podcast, making a commercial spot or jingle, or scoring a movie - you need a DAI to get your signals into and out of your DAW (whether it’s a Raspberry Pi or a JoeCo pro recorder). It can be as simple as a $35 2x2 USB box the size of a sardine tin or an 18x18 out prosumer level device with onboard mic preamps like a TASCAM US20-20 to a complete 24 channel pro quality mixer / preamp like the PreSonus StudioLive 64S with BT, DSP, and a host of sophisticated features. But you gotta have one, and it should be chosen for your intended uses.
For live recording of more than 2 tracks simultaneously, you really need a MC DAI. Over the years, I’ve often used more than 2 mics on direct stereo recording, largely because I didn’t have a MC DAI. When you do it this way, what you record is what you get – there’s no way to pan, apply DSP, or do anything else to the signal from an individual mic once the recording is made. Any post-processing is limited to the cumulative content of a channel. I just plug the stereo line outs into my M-Audio 2x2 DAI, set everything up as carefully as I can during the sound check, and pretend I’m back in the golden days of direct-to-vinyl recording. After all, this is how Sheffield got started (albeit with better stuff).
Here’s the little 6x2 Samson analog mixer I’ve used for several years for live recording with more than 2 mics, next to the M-Audio 2x2 USB DAI that’s been my home studio go-to since the day I got it:
I recommend the M-Audio line very highly - it’s built like a tank, sounds & feels great, looks new after 5+ years of being dragged to clubs, theaters, etc, and works great with a Pi as my mobile DAW. I also love the EVO line from Audient. The 4 and 8 are versatile, inexpensive (<$200), high quality DAIs with good preamps and convertors, great specs, excellent software for control & monitoring on PC or mobile, and a great feel (which is purely subjective but a valuable guide to how much you’ll use and enjoy something, if you’re anything like me). I’m thinking of switching to Audient EVO for my rig.
A 2x2 DAI is fine for a home studio in which no more than 2 tracks will be recorded simultaneously. I, like many musicians, build up virtual bands by adding each instrument alongside the already recorded parts until I have an ensemble in the can. If I want to add a bit of ambience, I put a pair of microphones in place and run them through separate channels of the analog mixer (panned to the best stage location) while my main instrument pickup output and/or mic(s) are in dedicated channels of their own.
Given the range of cost and complexity in equipment beloved and used by AS people, I’d be remiss if I ignored the upper reaches of the industry. There’s a lot of wonderful pro stuff out there If you want to go for it. Here’s a typical pro 64 track DAI / digital mixer with an integral USB DAI from PreSonus.
Few of us would set up a studio with devices like this one with a Raspberry Pi as the DAW. But even if you just want to have a pro level studio at home, having a Pi Daw as a backup device isn’t a bad idea. Given that many original recordings have vanished over the years, including the loss of one of the original master tapes for Kind of Blue, it’s prudent to have backup.
There are some very fine mics out there for recording. As an old codger, I’m still a fan of the Shure SM-57 (better for instruments) and 58 (better for vocals), both of which are long time industry standards and still available for under $100. My four have served me well for many years. There are many excellent and durable mics from Shure, AKG, Sennheiser, Rode, and many others in the $300 and under range that will let you capture live music with excellent fidelity. Money does buy you happiness in microphones – the really good ones are really good! But even a $100 Shure will make a pretty decent recording. If your live recordings made with mics like Shure SMs sound bad, it’s far more likely that it was your fault than the microphones’.
If you have mics you already bought for another purpose, try them before deciding you need more. I’ve been impressed with the SQ of live recordings of my acoustic guitars and piano made with my Dayton EMM-6 calibrated reference mic (hardly a studio standard). And when it became obvious that Covid was not going to abandon us for a long time, I bought an inexpensive set of wireless headset mics to use for live streaming (which is all we can do at the club until it becomes safe and sane to play to live audiences again). For under $100, I got 2 headset boom mics plus 2 lavalier mics, with a pair of wireless transmitters and a 2 channel receiver. I didn’t expect much, but I got value well beyond my wildest dreams. The wireless system works amazingly well, but these are actually pretty fine mics that can be plugged directly into a DAI. I made a mount that suspends one of the lavalier mics over the sound holes of my National tricone resonator guitar for a wonderfully rich and realistic capture. The headset mics are quite nice for vocals, and I’ve used them successfully sitting on floors and desktops as boundary mics for ambience capture. They’re worth their cost and far more, if you have any need for them (e.g. podcasting, video conferencing, Zoom parties etc).
You can use them for audio recording in your home studio. Although I’ve returned my share of low cost products that were marginal performers compared to the best of their breed, this particular Hotec wireless setup is truly dead silent and very resistant to RFI. The club in which I’m the house band leader is next to railroad tracks with serious power wires for the electrified Amtrak, regional rail and commuter trains that run on them. This mic system is even impervious to the interference from those wires that makes most electric instruments and devices on our club’s stage (including our DIs and even the house sound system at times) buzz like giant bees with a thyroid problem.
STEREO AND PAIRED MICROPHONES
For me, a good stereo mic is the easiest and most reliable way to make a decent recording of a live musical performance in an unfamiliar setting. Trying to use multiple microphones in an untried venue when you have little or no knowledge about or experience with it is a pig in a poke. So my choice is the $300 Audio-Technica AT2022 pictured below (or any of a number of similar units).
The $700 Rode NT4 is a similar design but a bit higher quality, with balanced outputs (unlike the AT above) and a bit more robust feel.
Another fine alternative to multiple stand-mounted traditional mics is a small matched pair of C2 condenser mics from Behringer for an astounding $70. They come with a tiny stand that’s very nice for use on location and works well on a flat mic stand made for desk use (pictured below carrying one of my mics).
These have balanced outputs (XLR) but (like all condenser mics) require a power source and do not have onboard batteries. So they require phantom power, which means using them with a preamp / ADC that provides this (e.g. Steinberg’s UR22c described and pictured above). Here’s a nice summary of several other small, stereo mic pairs good for remote recording.
Another approach to remote stereo mics comes from the video world. There are some nice, small, wearable stereo mics (starting with the $30 Sony ECMCS3 shown below left and running up the scale to better sound and quality at higher prices, e.g. the $75 Sony ECM DS70P below right) that can be used for discrete, unobtrusive recording of live performances with decent sound quality. Clip one on the front or top a hat for close to true binaural recording (if you don’t let your head get in the way). These tiny pieces are not easily noticed – but you have to keep them away from the sources of extraneous noise that make so many captures like this unlistenable (e.g. the sound of your two day beard growth as it rubs your collar when the mic is clipped to an adjacent lapel).
There are far better small stereo mics than the sub-$100 pieces sold primarily for use with video cameras. Consider the $350 Sennheiser MKE440 (below left), a stereo shotgun pair of condenser capsules with onboard battery. A shotgun mic is seriously directional, so it’s not going to capture ambiance or a broad sound stage – but it’s a good approach to recording solo instrumentalists from a distant seat and for use in very noisy locations. There are similar two-capsule / single housing mics that are cardioid, many of which are high quality like the $600 Rode VideoMic Pro (below right).
Some of these stereo mics for video use are capable to capturing good sound, but for audiophile quality you really do have to spend a lot more than it takes to get the same sound quality from a pair of small traditional mics. The best use for tiny video mics in my book is unobtrusive recording of programs for which you simply can’t use traditional mics. It takes some ingenuity and perhaps a bit of equipment compromise, e.g. none of these has balanced outputs and almost all have short nondetachable cables with 3,5mm TRS plugs on them.
If you like the idea of a tiny stereo mic for unobtrusive binaural recording of live performances with a Raspberry Pi, I strongly suggest starting out with a HiFiBerry DAC+ADC Pro. The sound quality’s more than decent, and it has an analog stereo 3.5mm input jack so you can plug a good video mic directly into it.
Microphones vary in output levels from as low as a millivolt to tenths of a volt, depending on design. All microphones are simple devices that generate an electrical potential in response to action against a suspended element, e.g. from the pressure of the air pulses in a sound wave (a dynamic mic) or the velocity of the air pulses (a ribbon mic). These are low output devices that require as much as 50 to 70 dB of gain to reach line level voltage.
Condenser mics use a conductive membrane held parallel to a conductive backplate so that the element forms a capacitor (which were called condensers when codgers like me were young). A stable charging DC voltage is placed across it by internal battery, external PS, or “phantom power” in the cable. Vibration of the movable membrane from sound pressure waves changes the capacitance of the element, which in turn alters the voltage flowing through the circuit. This creates an AC signal by modulating the applied voltage drop, and that delta is the audio signal. Condenser mics have high output, requiring as little as 10 to 30 dB of gain for any practical use, because they have a built-in “head amp” to set the output level.
USB mics have both preamplifiers and ADC circuitry built in. They vary in the nature of their elements & capsules, with dynamic and condenser versions readily available from the usual suspects. Although many are inexpensive and marginal for any use other than simple communication, there are some very good ones on the market that will do the job for even the most serious and professional podcaster. No USB mic I’ve ever seen or read about can compare to a top quality traditional mic for live music capture. But a few are quite good, e.g. the Blue Yeti X and the Shure MV7, and are well worth their cost for beginners. The Yeti has 4 capsules that can be set to record from a variety of patterns, while the Shure is purely a cardiod mic with electronics and ADC for a USB output as well as a balanced analog output into an XLR. The electronics are decent+ in both, and the Shure lets you use the USB and XLR outputs simultaneously.
HOW MUCH DIFFERENCE WILL A BETTER MIC & PREAMP MAKE ON A RASPBERRY PI?
Make no mistake about it – a Raspberry Pi 4 is quite capable of capturing truly excellent audio in live recording. If you’re at all serious about recording live music, you need decent mics and preamplification beyond even the best USB mics – the difference will be obvious, and it’s definitely worth spending $100+ per mic. The Shure SM57 and 58 are workhorses that have made many people very happy for many years. At under $100 each, they’re the absolute minimum I’d consider for capturing music well enough to enjoy listening to it. If you want a pair of decent mics to get started, head on over to one of the big box music stores and try out whatever’s on sale from good makers like Shure, AKG, Rode, Sennheiser, etc. Try house brands as well – some of the products by Samson etc are great value and do a decent job. Listen to your own voice through phones using each mic – it’s very revealing. I also find that listening to music played back on my phone through a mic and headset is quite helpful because the sound is already tinny, which reveals many character flaws not as evident with full spectrum program material.
The mic preamp in my M-Audio 2x2 DAI is the lowest level I’ve found acceptable. It’s quite capable although a bit limited compared to my “good stuff”, with slight veiling and a bit of transient blur. But it’s far from a painful compromise and sufficient for most beginners to learn both what they’re doing and where they want to take it before spending more money than may be necessary. I do have a few small and inexpensive mono mic preamps that are outstanding in SQ and value. My favorite is the ART Original (pictured below left), a fantastic little product that I’ve used for many years. It’s a tube device (12AX7a) with balanced and unbalanced outputs and great sonics – a tiny hint of warmth, great accuracy, and an amazingly quiet input noise level of -129 dBu (A weighted, balanced). BTW, it costs $70.
You can also get this quality and performance for use without an external DAI at little more cost ($150), if you’re only recording one track at a time. I recommend the Behringer MIC500 USB (pictured below right), which is almost identical in spec to the ART but also has some DSP and an onboard ADC with USB out. And you can find several equally good multichannel analog and USB mic preamps at similar cost per channel.
Whether you’re a podcaster, recording your local church choir, or capturing a band at your favorite club or auditorium, a preamp serves several functions. A good mic preamp will maintain more openness at higher gain with much lower noise than those in an entry level DAI. And most offer features not available in the onboard electronics of low end DAIs, like phase reversal, filters, pads etc. If you’re starting out, I’d spend a few more dollars on a DAI with better preamp(s), because the jump to the next level of quality is far more costly than another $50 to $100 and the return is less dramatic than the difference between a $100 2 channel DAI and a $250 one. For example, a Grace M101 is a very nice single channel mic preamp – but it’s about $750. So the above advice is the most practical I can offer from my own experience and in line with a value based approach to quality audio.
SUMMARY DO LIST FOR A PI 4 DAW
Set up a Raspberry Pi 4 with 4 gigs of RAM to run the 64 bit Raspberry Pi OS from a USB 3 SSD; make sure it has full access to whatever networked folder(s) you plan to use for files.
- Remove or disable all apps and processes on it that you don’t need for this project.
- Pick a DAI that meets your needs and budget (using the above as a guide).
- Install the DAW software of your choice. You can put Ardour, Audacity & LMMS on the same machine without problem. Just don’t try to run them simultaneously.
- Select mics for your needs and budget (using the above as a guide).
- If you plan to use more than 2 mics at a time, use the above guide to find a suitable one.
- Get decent but rugged headphones for monitoring. Shuffling all this stuff around whiole trying to make a good recording results in dropped equipment, kicked cords, slips, falls, and other mishaps that threaten your really good phones (and your body parts, but that’s only secondary).
- Make sure you have and have read (carefully!!) every manual for each piece of software you’ll use. There are Wikis and forums for everything, and many are very useful if you use your head and don’t believe everything you read there. If it doesn’t make sense, it’s probably wrong.
- Learn to use JACK. You can live without it, but it’s so useful that you really shouldn’t even try. Here’s a link to Ardour’s advice about JACK.
MAKING YOUR OWN RECORDINGS WITH A RASPBERRY PI OR OTHER SBC
There’s a basic package you need to assemble for virtually all recording uses. The elements are simple and there are many options for each one. First, you need a computer.
The central unit is the processing unit. I use and far prefer a Raspberry Pi 4 with 4 or more gigs of RAM running the latest 64 bit Raspberry Pi OS from a USB SSD (I use 250G). This will work fine in stock form for basic recording of live music with
a decent power supply (stock Raspberry Pi Foundation product works well)
- I’ve observed no audible improvement from linear or lab grade PS
- no unnecessary running processes or other competing CPU demands
- no real time DSP
no power parasites
- external power for necessary USB devices except external SSD
- basic cooling – heat sinks and passive cooling case at minimum
options to improve performance for demanding use (see my prior article for details)
- overclock CPU to 2100 (MUST use fan cooling to keep it alive and stable)
- add ZRAM
- optimize memory split between CPU and CPU
- a decent power supply (stock Raspberry Pi Foundation product works well)
- I haven’t tried it yet, but the new Raspberry Pi 400 appears to be another great device for this use. It has 4 gigs of RAM, a faster processor than the 4 (1.8), and an integral keyboard for easy use in session. It has a huge heat sink inside and runs much cooler than an un-cooled 4. The only drawbacks compared to a 4 are 3 USB ports instead of 4 and mini HDMI outs instead of standard.
- I recommend installing both Audacity and Ardour on the Pi, along with JACK
To get your audio signals into and out of the computer, you need
a source of signal to capture
you can record from line outputs (e/g/ the board for a house sound reinforcement system)
- this limits you to whatever the sound person has set up for mics, EQ, etc
- it may be the only way to get the performance you want
- if you can use your own microphones, do so
- you can record from line outputs (e/g/ the board for a house sound reinforcement system)
a signal path into the computer
for analog capture, you need an ADC (standalone or in a DAI) and any interconnection adapters
- you can also use USB microphones directly into the Pi
- for digital capture (e.g. from a digital mixing board or house network) you need a D-to-D interface or direct input for the source (e.g. USB)
- for analog capture, you need an ADC (standalone or in a DAI) and any interconnection adapters
a signal path out of the computer
- for real time archive and backup, you need a network connection &/or a USB peripheral
for monitoring and quality control, you need a DAC (standalone or in a DAI)
- The embedded DAC in the Raspberry Pi SoC can be used through the headphone out jack in an emergency – the SQ is marginal at best and not useful for critical monitoring
You’ll need enough storage with you to be sure you can bring home everything you captured, along with the data generated by whatever recording program you use. Wav files are not small. For reference, here are multiple formats of one 3 minute song in stereo with acoustic guitar, National resonator guitar, harmonica, and vocal. I wrote and recorded it on request for the 2019 Philly Blues Society compilation disc. The 70 MB wav file is my master, and the others contain the same content converted for different uses (ignore the name change – I wrote the tune and decided to change the name). There are two mp3 files, one a direct conversion and one hit hard with DSP that punches it up for the majority of listeners (who will download the mp3s from the Society website and listen on their phones).
To complete the picture, here’s the size of the master file that went into the album (remastered by a “real” recording engineer from my 68,497 KB file):
Understand that bigger files are not in and of themselves more demanding of processing power. Wav files require no compression or decompression at any stage, so they’re actually easier on the system than more heavily manipulated formats. You just have to put them somewhere, which can be challenging when recording remotely on a Raspberry Pi.
When recording live performances, you need enough storage to be sure you can hold the entire program – and there are a lot of data in a DAW session. The folder for a typical Ardour session of one tune recorded in stereo can easily top 1GB. Each of the above files is a digital archive of one simple 3 minute song. So a typical 90 minutes of concert performance will shove a lot of gigs into your storage media. You need to have adequate capacity for your recordings as they’re captured, along with the associated data generated by your recording software (which can be considerable – the Ardour project file for that 68k song above is about half a MB). And you need a real time backup system “just in case”, because Murphy’s also a recording engineer.
For home studio recording, I record to NAS or a USB SSD disc, using NAS backup plus a second level BU via real time synch program like FreeFileSync (my favorite – it works flawlessly and is essentially instantaneous) and online BU / storage. But when you’re on the road with a Raspberry Pi, you don’t have the luxury of your own network and a desktop full o’ stuff. So you’re limited to what you can carry, power, and keep safe. For up to an hour of program material, you’re probably fine with a USB3 SSD of at least 250G, assuming there’s nothing on it except your OS and programs you need for recording. But this is a bare minimum, and I strongly recommend 500GB or more.
For location recording, there are many excellent portable SSDs these days that will save your files (and your bacon!). I’m exploring wireless storage, both for real time capture and for live backup, with the new WD My Passport Wireless Pro that carries its own serious power (6400mAh) and feels bulletproof. So far, so good.
MY BASIC RIG FOR MOBILE RECORDING WITH A RASPBERRY PI (OR OTHER SBC)
- Raspberry Pi with good power supply, appropriate software, and cooling (FLIRC or fan case)
- externally powered DAI with appropriate formats and connectors for my peripherals
- a matched pair of small cardioid microphones (with exceptions for a few specific settings)
- lightweight headphones (I use BT now that WiFi and BT are separate from the Pi’s USB bus)
- storage devices in triplicate (extra SSDs, in my case)
a mounting / storage case, platform, board etc to keep everything organized and usable
- I generally use foam inserts in the smallest briefcase or gig bag that will contain all the hardware, making custom cutouts for each piece
- I carry my microphones in a separate small case to minimize the potential for a fall or other traumatic event to destroy too much of my rig at once
sufficient power for all pieces from as few separate pieces as possible
- Bring at least one extra PS for each one you need – they fail at the worst possible times!
I carry three powered USB hubs, each with at least 2 more ports than I need because I’ve experienced isolated single port failure on more than one hub and PS over the years
- I like the Anker 10 port 60 watt hub (below left) and the Auckey (below right)
- There are a few quiet and stable PSs for USB use, with a simple $13 Motorola SMPS wall wart and a similar HP unit (157-10157-00, sold as a “Touch Pad Charger”) performing very well in lab testing. Even the Apple charger performed very well for noise. Interestingly, counterfeit versions of the Apple and a few other brand names were incredibly noisy and unstable – avoid these at any cost!
- I tried a few “lab” USB power supplies like the Tekpower (below) sold for cellphone and other USB device repair and similar commercial use. I don’t hear any difference between recordings made with these and with the standard Raspberry Pi Foundation USB-C charger they sell for the 4. Of course, YMMV.
MAKING A RECORDING OF LIVE MUSIC
The process of recording live performance is beyond the scope of this article, both because it’s an involved process and because it’s more than enough material for one or more articles on its own. Read the manual for each DAW you consider, and do a web search to find a tutorial you like. Check out the online edition of Sound on Sound magazine – I’ve enjoyed it for many years.
Here are some good basic tutorials that will get you started:
- Fundamentals of live sound recording – a simple intro to recording a live performance
- Here’s a fantastic page of tutorials from Audient covering dozens of needs and issues
- How to make a studio recording track by track
- How to mix live recordings
- A pro’s approach to recording a vocalist with piano accompaniment
CAPTURING LIVE PERFORMANCES ON LOCATION
These links will point you in the right direction for use Ardour and Audacity:
- Your first recording with Audacity
- Getting started with Audacity (YouTube)
- Editing audio files with Audacity (YouTube)
- Recording the guitar with Audacity
- The Ardour manual
- Getting started with Ardour (a half hour YouTube video) The presenter is a bit awkward and not entirely correct about everything, e.g. why you might not want to monitor through Ardour (discussed above – his explanation is sometimes correct, but for us it’s because marginal CPUs can’t keep up with recording, processing, and monitoring). Overall, it’s a fine source for getting started.
- The FLOSS manual for Ardour
- Recording multiple parts one by one with Ardour (YouTube - terrible music but useful methodology!)
A BIT ABOUT RIPPING WITH A PI (THE SUBJECT OF ANOTHER ARTICLE)
For starters, you can rip CDs with a Pi just as easily and well as you can with a more serious computer. There are so many programs for ripping that I won’t bore you with a list you probably already know. You can rip CDs on a Pi with VLC, JRiver, Volumio, and a number of other similar packages. And there are a few interesting ripping programs created for the Pi, like abcde (A Better CD Encoder) and PiAutoRip. But VLC does the job very well. Unlike vinyl, CDs shouldn’t require any post processing – the ripped files should be sonic duplicates of the CD in any decent format (I use FLAC).
A Pi is as good for vinyl ripping as any other device, if you’re happy with Audacity and willing to put in the time and effort to get well tagged, properly named files for every track on an album. It doesn’t automate cleanup, tagging, track naming etc), so there’s a bit of manual labor involved. Still, it’s a great way to get top quality digital files of your vinyl if you’re willing to put in the time and effort. I’m working on a piece with great detail about vinyl ripping as well as audio file editing and manipulation. Enjoy every note and stay tuned for more info!