Being a musician wasn't easy in the 90s. It meant improvising with whatever junk you had lying around the room.
My old IBM PS/2 was almost useless for anything remotely musical. But I had high hopes for my AMD 386 DX-40 MHz, which, of course, didn't come equipped with an audio interface.
It did, however, have plenty of boring hardware, like a couple of serial ports, COM1 and COM2, and a parallel port, LPT1, normally used to connect my dot-matrix printer.
Serial ports send data one bit at a time. The parallel port, on the other hand, can send eight bits simultaneously, at a rate determined by the port's operating speed. Its original purpose was to feed data, checksums and control codes to a printer.
But people are weird, and some weird bloke had the unusual idea of sending raw PCM data to the parallel port. Each bit contributes a different amount to the output voltage: the more significant the bit, the greater its contribution. If you send the raw PCM data at the right speed, a simple and cheap resistor ladder is enough to build a Digital-to-Analog Converter (DAC) and turn those streams of bits into an actual audio signal.
I couldn't do any soldering, so I got a homemade audio interface from a friend.
It consisted of a 25-pin connector mounted on a small prototyping board, eight connections wired through a handful of resistors, a mini jack output, and a lot of optimism.
Along with a generous amount of static noise and zero respect for dynamic range, it could finally produce polyphonic sound, gloriously in mono.
Back then, MODPLAY was the go-to tool for playing modules written by someone else, downloaded from those early BBSs, and it worked perfectly with my rusty audio interface and my Sony compact Hi-Fi.
But if you wanted to create your own music, you needed a tracker like MODEDIT. That editor relied on an audio library called MODRES, from the same author as MODPLAY.
I was amazed, and of course I had to disassemble the MODRES library to learn how it worked on the inside. Later on, when I got my first email address at the university, I wrote to Ralf Brown and asked if the information I had gathered was interesting.
He said yes, and my name is still in the credits of the Interrupt List, which remains one of my favourite pieces of relaxing bedtime reading. Even Andrew Tanenbaum, whose books I was studying at the time, thanked Ralf in his acknowledgements. This means we have something in common!
From the tracker to tape
The LPT port let me play noisy 8-bit tracks with virtually no dynamic range. I could also write them using MODEDIT (and, later on, FastTracker II).
I have always found the tracker approach incredibly natural: a tracker forces you to look at the song as a whole, not just at a single moment. You can't just focus on the cool riff. You have to think about the entire structure: Does the chorus go here? Should the verse repeat twice or three times? Is the intro too long? Shall I remove this part entirely?
A tracker is a visual map of the song.
It also imposes very strict limits: a fixed number of channels, a fixed number of rows per pattern. For some people these boundaries are a nightmare, but for neurodivergent brains they are a relief. I never have to stare at an empty DAW and wonder where to start.
So the songs were somehow ready, and I wanted to hear how they really sounded. But the parallel port could only send data. There was no way to plug in and record a guitar or a microphone.
And, guess what? I didn't own a mixer, nor did I own a multitrack recorder at the time.
What I had was a mid-range Sony HST-D301 Hi-Fi with a cassette deck, stereo RCA inputs (normally used for a CD player) and a headphone out. Just in case, I also borrowed my sister's portable Aiwa cassette recorder, which wasn't great but had a stereo RCA input and a headphone out. Finally, I managed to get my hands on a handmade mixer built by a friend with a handful of resistors and some chips, featuring two stereo RCA inputs, one stereo RCA output, and absolutely nothing else.
As for the guitar, all I had was a basic 10-Watt Aria amplifier with no line out. I didn't give up. I faked a line out by connecting the speaker cables directly into the handmade mixer, keeping the volume as low as possible to avoid blowing everything up.
I was amazed at how the guitar sounded exactly like an early Beatles record: absolutely terrible. But at least now I could record the tracks along with the guitar, and I could choose whether to pan it hard left, hard right, or center.
And it was finally on tape.
Bounce Like Crazy
And what if I wanted to add more layers? A second guitar, a solo, a vocal track, some effects, or just more noise?
I could play back the existing recording, feed the mixer from the headphone output, and play along with it. Then I'd record the whole thing again, adding another layer to the mix.
This is called track bouncing, and it's exactly the kind of trick people used to squeeze more tracks out of a four-track recorder. You record several parts together onto a single track, freeing up the others for new overdubs. Of course, every bounce means another generation of tape, and a little more loss in sound quality.
The Beatles did it all the time. They had four tracks to work with, so they bounced like crazy.
Me, on the other hand, I only a stereo track, cheap and noisy hardware, and a lot of unreasonable optimism. So I started bouncing too.
The Mechanical Horror
But bouncing on a four-track machine is one thing. Bouncing between two consumer cassette decks is another kind of punishment.
The two machines were my Sony Hi-Fi and my sister's Aiwa, and their motors ran at slightly different speeds. If I played a tape on the Sony and recorded it on the Aiwa, the pitch drifted a little. On the next pass, going back the other way, it drifted again, and a little more each time.
So every bounce went like this: play the tape on the Sony, add the guitar through the mixer, record everything on the Aiwa. Then swap all the cables, play the Aiwa, add another layer, record it back on the Sony. And so on.
The result? Noisy, of course. But extremely magical.
Why? I was using the cheapest tools on the market, and still the record sounded better than expected.
I believe it was the tape itself. Every bounce added a layer of hiss, but tape also squashes the loudest peaks a little. The static noise of the 386, the hum of the mixer and the tape hiss somehow glued the instruments together. And the imperfections of the motors made the whole mix sway with a subtle, almost inaudible vibrato, a phenomenon known as wow and flutter.
Pink Floyd experimented a lot with tape. So did I, mostly by accident.
Years later, of course, I finally moved out of the Stone Age and bought a real sound card. The legendary Sound Blaster AWE32 was already off the market, so I settled for an AWE64, followed later by a Live!, an Audigy 2, and so on.
And in the late 90s I was one of the first to adopt Cakewalk Pro Audio, which ran proudly on a Pentium with a mechanical hard drive, and which survived Gibson's corporate fever dreams, years of being free, then paid, then subscription-locked, then (finally) free again.
Today I own professional guitars, great amps, microphones, high-end amp simulators and a DAW with more tracks than I will ever need.
But I just got myself a box full of resistors, audio chips, converters, sockets and a Raspberry Pi Pico, because I have finally learned how to solder.





Top comments (0)