Bending an Atari 2600 Pac Man game with <a href="http://buza.mitplw.com">Kyle Buza</a>'s "virtual circuit bending" <a href="http://cycling74.com/products/max/video-jitter/">Jitter</a> external, <a href="http://www.mmonoplayer.com/jit2600.php">jit.atari2600</a>. I just made a few trivial modifications to his demo patch and messed with a few things. His work is great, and very deep - I wish more people were using it (I'll be one soon).

Pac Man bending with Kyle Buza's jit.atari2600

Roland has been teasing their new AIRA line (which will include a TR-808 "remix," the TR-08), and it <a href="http://createdigitalmusic.com/2014/02/rolands-four-aira-instruments-now-know-lot-keyboard-component-analog-modeling/">seems the line will be based around a virtual analog</a>, featuring a <a href="https://www.youtube.com/watch?v=cy428RHu1mw#t=127">technology that Roland is calling "ACB"</a> (Analog Circuit Behavior). This is apparently a very polarizing issue, and lots of people have reacted negatively to the announcement - they think that a computer model of analog technology could never faithfully recreate the original.

I think that is totally incorrect. As part of my PhD studies at CCRMA (Stanford University's Center for Computer Research in Music and Acoustics), I've been researching the circuitry of the 808 for a year, and one of my findings has been that there is no reason that the 808 cannot be simulated with virtual analog.

My research focuses on breaking the circuit into functional blocks, and creating physically-informed analyses and digital models of each block. Even though my research won't be presented or published in conference proceedings until later this year, I wanted to share something that I believe is a strong counter example to the idea that analog devices cannot be well-simulated in a digital system.

I took the circuitry of the TR-808 bass drum sound generator and entered it manually into <a href="http://www.linear.com/designtools/software/">LTSpice</a>, a free <a href="http://en.wikipedia.org/wiki/SPICE">SPICE</a> ("Simulation Program with Integrated Circuit Emphasis", technology whose development began at Berkeley in the 70s). Setting the tone knob fully open, and the volume knob at 50%, I ran a transient analysis with a maximum time step of 0.000002267573 seconds (10x oversampled, with respect to CD audio, where fs=44100), exported the output voltage into Matlab, resampled to CD audio with Matlab's interp1 function, and wrote it out to a .wav file.

The first 12 seconds simulate the case where the decay knob is turned all the way up - the bass drum will ring out for a long time. The next 12 seconds simulate the case where the decay knob is turned down a bit. You can see and hear that the long decay case has all sorts of variation between notes, and the shorter decay case has much less variation. If we turn down the decay enough, there will be hardly any variation between notes.

I think that this clears up one misconception that many hold about analog drum machines, regarding the so-called "machine gun effect." The "machine gun effect" is the (usually un desirable) effect of playing many identical repeated drum hits. In the case of sample-based or signal-based models, people have created all sorts of workaround to deal with the effect, including randomizing synthesis parameters on each hit, or triggering from a large library of near- but not totally-identical samples.

Although it is usually attributed to the non-linearities of analog circuitry, or some sort of real-world randomness, the dominant reason that TR-808 bass drums (and by extension, other sound generators involving bridged-T networks or envelope generators) can sound different when many notes are played in quick succession is that a filter's response is going to be added to any response that is already happening (the filters obey superposition). The decaying tail of the previous note and the new note can interfere constructively, destructively, or anywhere in-between. Anyone who has jumped on a trampoline knows the phenomenon of "double bouncing" and the disappointment of a slightly mis-timed double bounce. This is precisely the same concept, and it has everything to do with the fact that it is a filter creating the drum sound, and nothing to do with the fact that that filter is analog.

This simulation actually runs much slower than real-time (it takes about a minute to render a second of audio). One point of my research (besides drawing attention to the ingenious designs of the 808's creators and supporting the work of drum machine modders and designers) is making models that are physically-informed, but simplified enough to run in real-time.

-Kurt (kwerner@ccrma.stanford.edu)

TR-808 bass drum simulated in SPICE

I'm working on designing an Arduino shield to work with old AY-3-8912 PSG chips. Here's the results of my fist tests interfacing the chip with the Arduino. The was based of Octobit's work on the AY-3-8910 (available at <a href="http://playground.arduino.cc/Main/AY38910#.UwWYkZE2zwK">http://playground.arduino.cc/Main/AY38910#.UwWYkZE2zwK</a>), which was based off of kalshagar's work on the related Yamaha YMZ294 (available at <a href="http://kalshagar.wikispaces.com/Arduino+and+a+YMZ294">http://kalshagar.wikispaces.com/Arduino+and+a+YMZ294</a>). I had to modify Octobit's work to accommodate the different pin arrangement of the AY-3-8912 and correct the wiring of the 2MHz clock (his diagram shows a Vcc line going to ground). I modified his starter code to write a few little tunes and play around with various aspects of the AY-3-8912's synthesis.

0:00-0:08 : channel A playing a dinky square wave melody with no amplitude envelope
0:08-0:24 : a classic PSG timbre trick. This series of chips can produce three channels that contain square waves, linear feedback shift register (LFSR) noise, or both. One global amplitude envelope system could be applied to all of the channels. To get around the timbre limitations of just square waves + noise, composers would apply the global amplitude envelope very fast to get crude but varied AM timbres. Here, I've applied this technique very inelegantly by applying a very fast triangle wave amplitude modulator at random frequencies which change every few notes. It can actually be used to great effect by controlling the modulation:carrier ratio to place sidebands carefully (just like in FM synthesis).
0:24-0:40 : The same trick, with the melody at a lower octave.
0:40-1:02 : Using channels A & B to play a Bach-ish one-to-one counterpoint, with a more normal global amplitude envelope.
1:02-1:12 : Same thing, sped up really fast to give a Galaga-style sound effect.
1:12-1:22 : Same thing, but with the original melody. This one actually sounds a lot like Galaga...
1:22-1:54 : The original Melody, but with noise enabled on channel A. Each note, the period of the LFSR changes, giving different noise timbres each time.

The AY-3-8912 was a popular 3-voice PSG chip from the 1980s, which was used in the Intellivision and Vectrex video game consoles, as well as various arcade games and home computers. Even early on in the chip’s history, it was popular for hobbyists and hackers (as shown by at least one magazine article on its use in the context of the Atari). Today, several people have built off of early projects and maintain ongoing discussions focused on using the AY-3-8912 with microcontrollers. The sound of the chip remains popular enough that is has even been digitally modelled as part of Chipsounds, a commercial digital audio workstation (DAW) plugin by Plogue Art et Technologie and MAME (the multiple arcade machine emulator).

AY-3-8912 + Arduino, First Tests

Another test of my AY-3-8912 + Arduino setup (you can hear the first one here: https://soundcloud.com/kurt-james-werner/ay-3-8912-arduino-first-tests). I've started experimenting with more of the chip's features, and wrote up a very simple semi-algorithmic tune. Octave offsets, noise timing, noise tuning, and envelope selection and period are all chosen by a simple algorithm, which combines randomness with fixed elements. For instance, a block of code like:

if ((i==0)||(i==8)) {
    write_data(0x06,31); // long noise period
} else if ((i==4)||(i==12)) {
    write_data(0x06,1); // short noise period
} else {
    write_data(0x06, random(0,31)); // random noise period
}

controls the period of the linear feedback shift register (LFSR) noise. On beats (represented by the variable i) 0 and 8 (of 16), the noise will have a long period and a short period, respectively. On other beats, the period is random. This is a crude nod to rock rhythms with downbeat bass drums and offbeat snares.

AY-3-8912 + Arduino, Second Test

A quick demo of my TR-808 bass drum physical model - normal behavior (various decay times and accent levels) and some simple bends (extra decay time via. modded feedback buffer, tuning bend, attack tuning bend). Tone knob is set at 80% throughout.

Soundcloud pulls some funny business with streaming encoding (I suspect its just transient smearing or time domain aliasing from sloppy perceptual audio coding... something they could have mitigated with block switching or explicit encoding of transients...). Since a lot of the sound of the 808 bass drum is bundled into the transient behavior, I recommend downloading it and akin a look/listen in your DAW of choice - you should get the un-messed-with .wav file I uploaded.

Demo of TR-808 Bass Drum Physical Model

An improvised piece from the second night of the 2014 Circuit Benders' Ball in Nashville, Tennessee. Featuring vocalist Chervon Koeune (<a href="http://www.chervonmusic.com">Cher Von</a>) of Louisville, Kentucky; <a href="https://ccrma.stanford.edu/~kwerner/">Kurt James Werner</a> (<a href="https://ccrma.stanford.edu">CCRMA</a>) of Stanford, California on iPhone circuit-bending emulator <a href="http://bent.fm">bent.fm</a> and mixer/EQ/FX sends; Arnie Holder (<a href="http://datathrash.bandcamp.com/album/hymnal">Abandoned on Fire</a> / <a href="http://datathrash.com">Datathrash Records</a>) of Bowling Green, Kentucky on modded Gameboy + LSDJ; and Dustin Jesch of Fargo, North Dakota on circuit-bent drum machine.

We got a nice review at Stephen Trageser's <a href="http://www.nashvillescene.com">Nashville Scene</a> article "<a href="http://www.nashvillescene.com/nashvillecream/archives/2014/04/14/recap-notes-on-circuit-benders-ball-2014">Recap: Notes on Circuit Benders' Ball 2014</a>."

pic: Stephen Trageser

[Cher Von, Kurt James Werner, Arnie Holder, & Dustin Jesch] @ Circuit Benders' Ball 2014

tl;dr—"Return of the Obra Dinn" is excellent, and I reworked part of its soundtrack as a strict 1-bit composition, mirroring its crisp, retro 1-bit graphics. Scientific description of some aspects in this paper: 
Working link: https://www.researchgate.net/publication/336915189_Generalizations_of_velvet_noise_and_their_use_in_1-bit_music
Original link (broken?): (http://dafx2019.bcu.ac.uk/papers/DAFx2019_paper_53.pdf)

"Return of the Obra Dinn" (2018) is a fantastic game released in 2018. Its creator, Lucas Pope, describes Obra Dinn as "An Insurance Adventure with Minimal Colour." The graphics are entirely 1-bit, meaning that each pixel of the screen takes on only one of two possible colors: light or dark (nominally white or black). This hearkens back to the visual display style of old computers with  monochrome monitors, like the Commodore 64 or Apple IIe, and gives the game a unique and beautiful aesthetic. To represent grayscale values, Pope uses a technique called spatial dithering. Cleverly, he actually deploys a variety of dithering algorithms (including sometimes suppressing the dithering entirely) to differentiate textures in the game.

As a cover and tribute, I've rearranged part of the game's original (rollicking, acoustic) soundtrack—part of the music from the "Unholy Captives" chapter—in a 1-bit *musical* style. This not only mirrors the 1-bit graphics of the game, but mimics the 1-bit "beeper" audio capabilities of many early computers like the Apple IIe and ZX Spectrum.

1-bit music poses a unique compositional challenge. First of all, there is no signal amplitude control, making volume control difficult. Furthermore, in the 1-bit domain, addition of 1-bit signals does not properly exist. This is because certain values can be added (0+0=0, 0+1=1, 1+0=1), but the operation 1+1=2 cannot, since "2" is forbidden! Strictly speaking, this limits polyphony to one sound. Working around these limitations to produce music that is not monotonous and can express chords, layering of multiple instruments, and convincing audio effects becomes quite the task.

In this track, I manage to coax up to 6 layers out of the sound at once: bass, a three-note chord, a melody, and percussion. Part of the trick is just interlocking parts in the arrangements—here, for instance, the bass and chords rarely play at the same time. But in general, there are moments where up to 6 sounds have to be expressed with only 1 bit. The layers are mixed using a custom XOR-plus-error-feedback 1-bit mixing engine, combined with judicious use of thin pulse widths, custom 1-bit wavetable synths, and a special kind of sparse noise that can be XOR'd with signals without overwhelming them. Furthermore, the melody employs what I think may be a novel feedback delay effect that operates entirely in the 1-bit domain to give a sense of space to the track. This is mostly a note-for-note cover of Lucas Pope's original—involving elaborate 1-bit trickery in the sound synthesis, arrangement, and audio effects—but I did take one small liberty: an extra half measure of percussion added just before the start of the final phrase at 0:39.

The special sparse noise sequence is called "crushed velvet noise" and is described in my 2019 DAFx paper "Generalizations of velvet noise and their use in 1-bit music," available here: http://dafx2019.bcu.ac.uk/papers/DAFx2019_paper_53.pdf.

I've deviated from a strict 1-bit construction in just one way—The entire track was processed through Plogue's Apple IIe impulse response to make it sound like it is coming out of a real Apple IIe speaker, rather than being a "raw" stream of zeros and ones.

"Unholy Captives" from "Return of the Obra Dinn" (Lucas Pope, 2018): a 1-bit cover/tribute

Dr. Kurt James Werner is a Lecturer in Audio at the Sonic Arts Research Centre (SARC) of Queen's University Belfast, where he joined the faculty of Arts, English, and Languages in early 2017. As a researcher, he studies theoretical aspects of Wave Digital Filters and other virtual analog topics, computer modeling of circuit-bent instruments, and the history of music technology. As part of his Ph.D. in Computer-Based Music Theory and Acoustics from Stanford University's Center for Computer Research in Music and Acoustics (CCRMA), he wrote a doctoral dissertation entitled “Virtual Analog Modeling of Audio Circuitry Using Wave Digital Filters.” This proposed a number of new techniques for modeling audio circuitry, greatly expanding the class of circuits that can be modeled using the Wave Digital Filter approach to include circuits with complicated topologies and multiple nonlinear electrical elements. He composes electro-acoustic/acousmatic music including elements of 1-bit, chiptunes, musique concrète, circuit bending, algorithmic/generative composition, modular synthesis, and breakbeat.

Kurt James Werner

Popular tracks