What does Suno mean for us producers & musicians?

 I've been thinking about this for a month or two now, and there is still a lot to process (obviously).  In a way for us musicians and producers, it feels like a "pencils down" moment... The teacher looks at the clock, stops everyone, and we all have to hand in our work, wherever we are at.  We either wrote good/recorded songs that made a splash, or we are now going to drown in the sea of auto-produced stuff that people are about to unleash on the world of music consumption.  The reality we are about to witness is one where disposable songs are literally created on the spot by an app for an ephemeral job posting, a birthday wish, or tomorrow morning's alarm clock sound... And then maybe never played or heard again.  The value of a song is on a different level altogether, one that is hard to relate to... And it seems unlikely we will ever live again in a world where so-called "classics" are made, or at least where everyone agrees on them as such.  The quest of breaking the barriers of producing great music, something that I have literally been training half my life to do, is over.  The barrier doesn't exist anymore, for anyone. 

That's kind of a bleak outlook.  But there's another side to this.  The fact that music making will now be a fully democratized commodity actually secretly elevates some particular aspects of music making. Consider the fact that making music without AI (especially weird stuff that is really outside the bounds of what AI is good at), will now be at a premium.  Humans will remember, and even seek out stuff that exists outside the norm.  We can think of the New Wave movement, in the early-mid 80's, as being sort of an analogous shift.  The LinnDrum was invented, MIDI, tons of studio automations... All of a sudden, there was a new sound and it was extremely low cost / high margin.  The big labels pushed it hard and it found its way to saturation, which led to... The punk music movement. This direct response was the antithesis of all that. It rewarded grass roots, real instruments, garage sound, edgy lyrics, people working their asses off without a $1M advance, just for the love of the craft.

If you carry that out, it's believable that people are still going to look toward bands and songs that are outside AI in some way.  I'm not trying to be a luddite here - being an AI collaborative band is something that will be very interesting (and, as far as I know, still yet to be any sort of a Zeitgeist), but the thing people will latch onto is: "this doesn't sound like AI!"...  Specifically, once everyone has access to make and (knowingly) hear generated music, it will immediately follow that real people, w/ real instruments, shot on real video cameras - making real music that is catchy and says something meaningful - will be held in higher regard because of the scarcity laws.  Mistakes, imperfections, even un-exciting, monotonous aspects of a song being allowed... flies in the face of everything I have seen in generated music thus far.  It will stick out like a sore dick in a whorehouse, and "no news is bad news".  

I really like the idea of the hybrid approach.  Write music from the heart, use AI to complement your skill set, create something that transcends, and be transparent about how you did it.  People love to know the process.  One of the things I have enjoyed the most about my friends playing with Suno is when all of the inputs are disclosed ("I made a recording this way, used a click, slider 1 at 25%, added style cues of blues + 4 on the floor backbeat, added lyric corrections", etc.).  The only thing greater than people's hate for being deceived is their love for being brought along for the ride of making a song. This is how some bands born on Youtube have gotten to be so big.  They don't just make the music, they take you on the journey and share the human experience.  At its core, making music is a very human experience, and that doesn't have to stop just because someone involved AI.  I would love to watch a video where someone makes a great record using elegant, judicious applications of AI within the process.  It's fair to say that, within our combined skills, we have decent musicianship, production, and to various extents, recording and mixing, but we lack in some areas like arrangement and repeatability.  If we use AI to fill in that gap, but show how we used it to make a great song, then learn and play it for people and bring them joy, that's a win.

So far this year, the stock market seems to be headed towards rewarding companies that have defensibility against AI takeover, and I think this theme dovetails perfectly within the music regime.

 

Building the ChordCaster Stomp: A Peek Inside My New Harmonizer Pedal

Every once in a while, a project grabs you by the collar and refuses to let go. For me, that project has been the ChordCaster Stomp - a compact, musician‑friendly harmonizer pedal built around a Teensy microcontroller, a custom audio front end, and a handful of ideas that have been rattling around in my head for years.

I wanted something that didn’t exist yet: a pedal that could grab a single note, understand the musical context, and generate harmony that feels intentional rather than robotic. Something that could switch between diatonic intelligence and unapologetically synthetic fixed‑interval parallel harmony. Something that guitarists, synth players, and experimentalists could all use without needing a theory textbook.

So I built it.

Here's a very quick & low effort demo of it in action. I plan to put together a full tour at a later date.

The Hardware: A Custom Brain in a Custom Box

The first prototypes ran on a Teensy 4.1 with the standard audio board - great for development, but not something I’d ever want to ship. Once the firmware stabilized, I moved everything over to a Teensy 4.0 and started designing a custom audio board to match the needs of a real pedal.

The stock Teensy audio board is a solid reference, but it’s built for general‑purpose use, not guitar. I needed a proper high‑impedance input buffer so the pedal behaves like a real instrument input, not a line‑level consumer device. I needed a cleaner, more robust line output that plays nicely with amps, interfaces, and other pedals. And I needed a compact layout that fits inside a stompbox enclosure without a rat’s nest of jumper wires.

So I kept the heart of the Teensy audio board - the SGTL5000 codec - and rebuilt everything around it. The digital side stayed almost identical: I²S for audio, I²C for control, clean 3.3V rails, and the usual decoupling. But the analog side is all mine: a JFET input buffer, proper AC coupling, a stable output driver, and ESD protection so the pedal doesn’t fry itself the first time someone plugs in a cable on a static‑heavy stage.

The USB port is panel‑mounted, too. No more dangling micro‑USB connectors waiting to snap off.

Harmony That Feels Musical

The ChordCaster Stomp has one job: turn a single note into something bigger. Hit the footswitch while you are playing a single note - the pedal “grabs” the note you’re playing and holds the harmony until you release. It’s like a freeze pedal, but for harmony. Great for ambient swells, pads, or turning a single note into a chordal drone. That's the basics, but there are many variations available, which is sort of where the fun begins:

Diatonic Mode

This is the “intelligent” mode. You pick a key and a musical mode (major, minor, modal mixture), step on a footswitch, and the pedal chooses the correct harmony interval based on the scale. It’s the kind of harmony that follows you around like a well‑trained bandmate - always in key, always musical, but keyed off only the notes you want it to pay attention to.

Fixed‑Interval Mode

This one is pure attitude.

Instead of following the key, the pedal applies the same interval no matter what you play. Want everything harmonized a major third up? Done. Want parallel fifths like a medieval monk with a distortion pedal? Go for it. Want to stack weird intervals and make synth‑like textures? Absolutely.

This mode ignores theory and just does what you tell it to do.

Arpeggiator Mode

This is the mode I didn’t realize I needed until I built it. Instead of playing all the harmony voices at once, the pedal cycles through them one at a time, turning a single note into a rhythmic, evolving pattern. The tempo is fully adjustable: you can dial it in through the menu or simply tap the second footswitch to set it on the fly. It’s a surprisingly expressive feature - great for ambient pulses, sequencer‑like textures, or adding motion to otherwise static chords. It feels less like a harmonizer and more like a tiny, obedient arpeggiator living under your foot.

Other Features

Beyond that, it features multiple outputs, reverb, FX and lots of other goodies.

The Idea: A Pedal That Understands Musicians

I’ve always loved harmonizers, but most of them fall into two camps:

1. Too simple - fixed intervals only, no musical awareness

2. Too complicated - menus, presets, MIDI, and a learning curve that scares off half the people who might enjoy them

The ChordCaster Stomp sits right in the middle. It’s musical, but not fussy. It’s expressive, but not overwhelming. It’s smart, but not bossy.

You turn the knob, pick your mode, and play. The pedal does the rest.

Where It’s Going Next

Right now, I’m refining the PCB layout, tightening up the analog noise floor, and polishing the UI so it feels natural on stage. The firmware is already solid - the Teensy 4.0 has more than enough horsepower for real‑time harmony, and the 360K of flash I’m using fits comfortably.

Once the hardware is locked in, I’ll start thinking about enclosures, graphics, and maybe even a small production run (but no promises).

For now, I’m just enjoying the process of turning an idea into a real, playable instrument.

If you’ve ever wanted a pedal that understands harmony the way you do - or the way you wish your bandmates did - the ChordCaster Stomp might be exactly what you’ve been missing.

The software is available on my Github, if you want to "check it out" (haha)

An Alternative Take on AI Doom and Gloom

 I've purposely held my tongue until now on commenting about "AI" (or, more specifically as has come to be known, GAN or Generative Adversarial Networks).  It seems like it is very in-style to complain about how it has made a real mess of things, it is displacing jobs, the product it creates lacks soul, it's going to get smart and kill us all, etc. etc.  But I'm not here to do any of that. Rather I am going to remind everyone of how amazing a phenomenon it is to watch a disruptive technology becoming democratized

From the time of its (seeming) introduction to the public at large, around November of 2022, to late 2023, the growth and adoption rate has been nothing short of explosive. It features the fastest adoption rate of any new technology ever, by a broad margin.  To give a reference, the adoption rate for AI image and text generation, real-world uses, in just 12 months is comparable to all of that of the another disruptive technology, the World Wide Web, taking place between 1995 and 2001 (6 years), and smartphones between 2009 and 2013 (3 years)*.  It seems like we are seeing a halving approximately every 10-15 years.

The mind blowing thing about this is, it would not be crazy to expect that the next similarly disruptive technology (presumably genuine artificial intelligence, known as Artificial General Intelligence, or AGI), to happen in the next 8 years (10 years from 2022), and will have an adoption rate of as little as 6 months.  

Yes, the current state of affairs is pretty ugly.  But so was a Tripod or Geocities webpage in 1995:

Side note, If you like that, be sure to check out the Geocities Archive for more fun snapshots of the mid 90's.

Back to my original point. When something new and shiny comes along, everyone wants to play around with it according to their own ideas of how to make it work for them, but it's not until the technology (and our mindset) matures a little, and an infrastructure is built.  In this case, HTML and web servers, and search came a long way before the World Wide Web was as usable as it was even in 2001.  Right now it seems that people think, for example, a good use of generative AI is making photos of Tom Hanks on acid in Walmart.  

Or replacing every business website with AI generated blurbs, featuring keywords they want to own in Google SERP.  Sure, the internet is going to be ugly, or even debateably, broken for a while.  But we haven't yet started to see a systemic infrastructure that bolsters the real power of this technology, and brings it to the users. 

Ultimately, will it be a great thing for humankind? Obviously, that is yet to be seen.  But I am hopeful and excited, if for nothing else than having a front-row seat.  

*Summary based on statistics obtained from PEW research and my own conclusions which admittedly may have been influenced or skewed by my personal experience

The Hellscape that is Google’s Web in 2023

Alternate title: "were we better off in 2015 2007?"

Time now for another anti-capitalist, “get off my lawn” posting for all the folks out there who won’t see it anyway, because they don’t read real blogs for the reasons specified in this very article.

The web has existed for 30 years now. One would think our ability to access information on it would keep getting better. However, I watch as web search is instead devolving every year, to the point where people are giving up and hoping for the next thing.  While this sounds dire, this kind of behavioral change has historical precedent. Remember running your own mail or web server, or better yet, having a phone that you might actually answer calls to, even if you don’t recognize the caller’s number?  Yes, those ideas are gone too. It's all thanks to the uncontrolled thirst for advertising.

Let’s walk through the experience of someone doing a simple Google search for “how to control poison ivy”.  The desired outcome would be to find a great testimony of someone who dealt with this personally, who goes on to procedurally list the approach(es) that worked for them.  

What you actually get: 

• The first result will be some company peddling their weed control poison 

• Followed by a bunch of weird “People also ask”  items with dropdowns that may or may not provide the answer to your actual pursuit. 

• Further down the SERP, you are immersed in a mine of “sponsored” pages with AI generated articles. Unless you fancy yourself a prospector, you can most likely simply discard the first 5 or so results in the Google SERP, which are “sponsored” garbage.

• By the time you hit “bottom” (wherever that is), you have lost hope in finding anything from a real person without some kind of bias.

Now, let’s say you, dear victim of post-modern web search, find and click on a promising result with an abstract showing, what appears to be, an answer to your question.  Buckle up, my friend.  Here’s what awaits you:

1. Site cookie acceptance dialog

Ahh yes, thank you so much GPDR, for making sure my privacy is respected.  Every single site uses third-party cookies to track my data for, you guessed it, more advertising. But like Batman, the GPDR is here to ensure the safety of Europeans (and by extension, the world) by enforcing all to click on a full page modal dialog (note: created by the very site designer who wants to track you).  You get to choose:  “Accept cookies”, “Leave Site”  (AKA “Get bent, content is not available if we can’t track you”), or sometimes most favorably, “view without accepting cookies”. When you select this option, the site may come up in its entirety and not track you! At least… not using cookies.  If you trust them.  People who just want to read a couple of sentences of content, without reading and thinking for a whole minute, are trained to simply click “accept”.

2. Great, you’re on the page.  Wait, here’s a popup in your face about cookies!

Do you want the site to send you notifications? Goodness lord, do NOT click yes on this. The result is pretty much akin to installing a malware taskbar into your browser in 2009.

3. Okay, you are reading some of the article.  Every blog-type article that ranks in the SERP seems to necessarily start with 2-3 paragraphs of awkward and formulaic cursory BS about why people search for what you searched for, all the things that happened when they tried to fix it the wrong way, etc… I’m guessing, to satisfy some unwieldy SEO requirement.  Scroll, scroll scroll!

4. Been here for 10 seconds? Decide whether you want to sign up for our newsletter! Again, a modal dialog box that you can’t look past. Disclose your email or (find a way to) close it.

5. Now… This article is actually getting good. You are close to getting real information, you can feel it!  Oh, hey, wait, it’s: Videos That Start Automatically!  Like Space Invaders, videos and other popup ads descend on you with increasing aggression over the next 30 seconds or so, narrowing your view inside an already minuscule viewport (assuming you had the audacity to do this on a phone). These abominations turn your quest to find urgent information about poison ivy into a horrible game of Whack A Mole. But harder than that, as you have to find and then tap a tiny (and intentionally obscured) ‘x’, in order to close the video/ad popup that obscures what you are trying to read.

6. Read a single glorious paragraph, and you will be rewarded with a chunk of random affiliate links and ads.   Was that the whole article? Is this the end?  Oh wait, no! There's another paragraph.  These places make Geocities and Tripod websites of the early 2000s look organized and well presented. It’s certainly reminiscent of the unanimously dark period of the WWW, when everyone was implementing their site with frames - certainly at least as offensive, if not more for its brazenness.  And we are not talking about hokey tabloid sites here, this is happening on trustworthy local news stations, local forums and bulletin websites, even school websites, trying to compensate for their insatiable SEO and web hosting costs.

What happened to the authentic articles and blog posts, written by the lowly peer web denizen of yore?  They are all but gone, as Google search algorithm continues to raise the bar for such simple sites to be included in the SERP with any reasonable ranking…  Leaving only the most insidious ones, created by a team of corporate content creators who are backed by a department full of SEO engineers who only have one job: to make this page go to the top. Yes,  a SERP for a given keyword set contains only one top result, and getting to the top is a zero-sum game.  With heavy competition for SERP dominance, it's a race to the bottom as companies copy each others' successful approaches, using them to shill their wares by sprinkling misinformation and links that serve only to redirect honest visitors.  Google Panda (2011, 2021) was, ironically intended to address over-optimized, ad-ridden sites, but instead appears only to have resulted in new spins of canned websites in which ads are carefully disguised as content… At least, in ways that are not outwardly apparent to a dumb web crawler.

It is painful, although not entirely unexpectable, the degree to which Google is a willing participant in this parade. They have answered the fierce call by shareholders for revenue growth by turning the SERP into a 10 ring circus. Searches that once yielded several page results at the top, with useful summaries and first hand authorship, now look as bad as any social media site - as videos, news items, shopping links, and irrelevant “People also ask” Q&A style items (ostensibly based on others popular searches, but triggered by a single stupid keyword in your search). These oddities appear in our faces, and feature sales-driven hyperbole and propaganda that panders to the most vulnerable among us. Every click feeds the bottom line of these advertisers, and in doing so, strengthens the machine to further pollute our attempts to gather useful and simple facts.

To what refuge have people turned as this debacle has unfolded?

Any refuge in this trying environment has experienced its ups and downs.  Take Reddit.com, one website where users developed and willingly invested their knowledge, and created a brain trust. Now reddit is slapping them in the face, by asserting uncompromising control and issuing a litany of new restrictions based on ensuring that information on Reddit is used only for their benefit, which boils down to ad-driven revenue.

High performance AI chatbots have recently surfaced, and people are desperately flocking to them as an alternative.  What’s not to like? Finally you can again ask a question and get a straight answer, (ostensibly) without being bombarded by ads! Even if it’s wrong, or based on a biased point of view.  These nascent tools have been rushed to the people before any effective ethical guardrails can be put in place.  It’s difficult if not impossible to know the source, or whether they are correct. The whole thing subsists on datasets that take massive liberties on  privacy and intellectual property rights. What’s worse, it’s a safe bet that If not now, very soon, the responses will be just another advertising circus.

What is there to do?  

Get back to the fields!  Start your own website! Write that blog post.  Keep content alive, and don’t let the web consist solely of advertising.  Let's form a web ring, like it’s 1997!  A new kind of search engine is needed, one that looks at the web a different way.  One that isn’t driven solely by advertising.  One that doesn’t squander the quality of its performance for the sake of shareholder value.  Bring back community-driven, crowd-sourced spirit. Wikipedia has done it with the world’s most comprehensive encyclopedia.  Let’s do something like that with web search!

GWSMO and Outlook Version 2303: Can't select "From" Addresses When Composing New Message

I found a bug/incompatibility with GWSMO (Google Workspace Sync for Microsoft Outlook, the Google Workspace sync integration for Outlook).  Here's what I found:

Steps to reproduce:

• Install Outlook 365 Version 2010 (I'm using click-to-run build Build 15726.20202)
• Test with a Google user that has multiple Send Mail As addresses configured properly under "Accounts and Import" tab in Gmail settings.
• Install GWSMO from .EXE using recommended settings. Follow the steps to authorize with your Google account via OAuth/browser
• Allow at least 5 minutes after everything is installed for GWSMO synchronization to progress far enough for Google account settings to propagate to Outlook
• Click on "New Message" in Outlook
• Click the From: dropdown to select one of the alternative send addresses
• Observe that the list of available addresses matches the list of Send Mail As addresses in Gmail settings

• Close Outlook, go into Windows Control Panel->Mail and remove the profile.
• Remove Outlook from the system
• Install Outlook Version 2303 (build 16227.20004)
• Repeat above steps to install GWSMO and configure Outlook, allow to sync, start a new Email, and click the From: dropdown
• Observe that now the list of available addresses does not appear when you click From:

Google believes this to be a UI issue with Microsoft Outlook, and thus would not look into it.  They told me to let Microsoft know about it, but I have a pretty good idea what they are going to say. Either way I didn't see anything about this issue anywhere on the web, so I decided to post about it.

So Much To Know...

I once spent an inordinate amount of time with a friend of mine, who I consider to be a master of recording & sound engineering.  He has been at it for decades, paid his dues, and worked with some greats.  He executes his craft with the utmost care and caring... (something I have come to realize is what separates the true pros from the learned hacks). Anyway, I remember watching him in do his thing in the studio as I assumed the role of a quiet observer.  Despite his obvious adeptness and ingenuity, he would always play the part of the  Absent-Minded Professor, as opposed to the James Bond smooth-operator type.  I think it was partly a schtick to make people feel more at ease, but there was a genuineness and willingness to be vulnerable, as though allowing himself to be human probably made the job less fatiguing.   I distinctly remember my favorite phrase of his.  On a couple of occasions, while turning knobs and fixing some problem, in a wonderous voice he would declare, "so much to know..." 

It's stuck with me because it's true about so many things, no matter how much of a master of your craft you are - there is always more to know.  So much more.  I've worked in IT and done various software development for about 30 years now, and yet that's been the theme for me lately - as I try to wrap my head around newer (to me) concepts such as containerization, full stack application development, and more.  I should have started learning these things about 10 years ago, so I have to play catch up.  But it's enough to make me feel old. 

Anyway, follow along as I may post some perspectives on (but not limited to) the following concepts, as I learn them:

• React/Mongoose/MongoDB
• Python/Flask/PyMongo
• Docker
• Kubernetes Clusters, K3S for high availability scalability, management thereof
• Tying all this together: MicroSaaS development and deployment

RANT TIME: Why do replies to a message I sent go to my spam folder?

Despite what one would think/hope, sending a message to a given address does not inherently give Google a high confidence that a reply from this address is expected (and, for example, that it should bypass spam checks). I have confirmed with Google's tech support that there is no way to automatically have this happen. The user can do the following:

1. Add the address to your contacts list in Gmail.

2. Check spam folder for replies, and mark it as "not spam" if something ends up there, which should influence the fate of future replies received. I can also approve an address at the domain level, i.e. if it is a big vendor or similar. I've had to do this with several of our Chinese vendors. I regularly ask engineering and purchasing to give me a list of the supplies we deal with, so I can approve them as a preventative measure.

For what it's worth, all of the false positive instances of reply -> spam we have experienced have involved the sender's email server having a problem. In the most recent case, it appears that the sender IP address appears on at least two internet blacklists. Since that is beyond my control, but we are trying to do business with these people, I can only add the domain to the approved senders list so that future replies from them should bypass the checks. However, if another company with a problematic email server replies to one of us, their message could very well still end up being marked spam.

Since Google can't help us, I am trying to figure out some kind of human process to defend against this, but to getting to a 0% false positive rate looks kind of ugly.  One idea I had is to make a script that is invoked when a user sends a message, and somehow adds the recipient address  to their contacts and/or some sort of approved sender list. 

Has anyone done this?