IBM PC Ramblings

(formerly the original Oldskool Beat)

Here's a handy menu to the various sections of this document:

Introduction

I love PC computer sound and music. It takes a lot of effort and patience to like PC music, to be honest, since every time you actually enjoy a warbly arpeggiated simulated chord through the speaker, or a cheezy-sounding Adlib FM soundtrack, or a snippet of 1-bit digitized sound forced through the PC speaker like a grape through a straw, people look at you funny. Real funny. I have great respect for musicians who wrote music (and programmers who wrote playback code) for PCs. It's an art to create a good soundtrack given the medium you have to work with.

And just what was medium? Specifically, what the output like? In the beginning, all you had was the internal PC speaker, which was a simple tone generator (real simple) without any voice or dynamics control; it could sustain a frequency and that was it. In the mid 80's, some better devices found their way into the hands of consumers. Some were extremely well-supported, like the PCjr/Tandy 3-voice sound chip and the Adlib music synthesizer card.

I created this section of Oldskool.org in the hopes that people will not only learn about the many different techniques and devices that software developers used to create music on early PCs, but also take a closer look at the sound/music itself and maybe appreciate it a bit more.

Sound Devices and Techniques

(The following is currently reprinted from Life Before Demos, but will be rewritten and expanded upon at a later date. I am ashamed to admit this shortcut, but I needed something here as a placeholder because I was running behind schedule for the opening day of Oldskool.org. If you've already read this in Life Before Demos, skip ahead to the Taste Test, since new information begins there.)

1981-1982

The PC speaker, driven by a chip that could only produce a simple tone at a fixed volume, was the only thing that kept us company. If it weren't for BASICA, we'd live in silence. BASICA had a PLAY statement that took real notes and octaves; you could bang out a melody relatively quickly, although it was loud and harsh. You could fake a chord by quickly alternating between different notes at the same time (an arpeggio), but this sounded artificial and bubbly. (If you didn't have a love for computer music, it would quickly drive you crazy.) Pianoman by Neil J. Rubenking was a music composition program that did this; you could compose each voice separately, and then combine them into an arpeggio. A gentler trick was to adjust the pitch up and down very finely, simulating vibrato. One voice, but at least it wasn't so harsh.

1983

The PCjr is released, and Tandy follows suit a year later with the Tandy 1000, which was a clone of the PCjr. One of the enhancements in the PCjr was the addition of a 3-voice sound chip that gave multiple channels, noise generation, tone envelopes, and volume control to the built-in speaker. Now we had something to play with. The BASICA that came with the PCjr and Tandy supported a 3-voice PLAY statement, which, if you played your cards right, could produce some fairly nice sound. One thing I discovered was that the Tandy chip had a hidden strength in low chord layering.

1984

Music Construction Set, programmed by Will Harvey, came out for the PC in 1984 from Electronic Arts. It had a real staff, with treble and bass clefs, and had a neato "construction set" motif--you could drag'n'drop notes onto the staff before "drag'n'drop" was a common catch phrase. Best of all, not only did it support the native sound chip of the PCjr/Tandy, but it could play four voices through the normal built-in speaker! (Granted, it was difficult to discern between the voices, but it was possible to hear the overall chord you were going for.) You could even print out the staff on your printer, although it was one long staff down the side of the page, and not nicely formatted sheet music. :-)

1986

Mindscape publishes Bank Street Music Writer, the first program I ever bought that came with its own hardware if you didn't own a Tandy or PCjr. The "Mindscape Music Board" was a 6 voice sound card which turned out to be a sine or square wave generator with simple attack, sustain, and delay parameters. Not exactly FastTracker 2 envelopes, but it was a start. :-) Plus, it attempted to print out real sheet music, and you could follow your voices on-screen as they played. I went nuts with this board, sometimes spending hours arranging the tunes my school choir was practicing. Although it was very good at producing solid chords (it was a tone generator, right?), it never took off, because the price was a bit high ($110) and it did sound a bit... "plinky". (Come to think of it, Music Construction Set for the Apple supported a similar board called the Mockingboard, but that never took off either.)

I'm fairly certain that I saw the Covox Speech Thing around this time as well. The Speech Thing was a simple digital-to-analog converter that you could connect to your parallel port to hear digitized sound. It sold for about $70, even though the parts cost about $15--including the speaker. :-)

1987

Adlib. :-) This famous board used a chip from Yamaha that produced Frequency-Modulated (FM) sound synthesis through 2 operators and a variety of parameters. You could utilize 9 melodic voices, or 6 melodic and 5 percussion. Armed with the odd Visual Composer, you could compose on a piano roll instead of a musical staff. It wasn't bad at all; in fact, it sounded pretty damn good. If programmed correctly, it could layer voices well, produce decent bass, and fairly full sounds. I still believe that the Adlib was (and still is) underused by the majority of people who composed for it.

While I didn't purchase my Adlib until 1989, the board was actually selling in 1987, and games started supporting it in 1988. Taito's arcade conversions done by Banana Development supported it passably, but it wasn't until 1990 that I heard simply beautiful music through it from a game called Continuum from Infogramme. The game consisted of jumping from platform to platform to reach a certain object, but the music was so good that I booted it up just for the music. (It also supported the Tandy sound chip, but since the music was composed for the Adlib, it was nowhere near the same quality.)

The great former C64 demogroup Vibrants also did some excellent music composed specifically for the Adlib board, but this wasn't until much later, in 1993, when they composed music for a few games. Their composition program, Edlib, is still freely available. Ever hear techno on an Adlib? :-) (Their true strength was jazz, which is what they usually composed.)

IBM Music Feature Card. This board was released from IBM in 1987 in an effort to draw MIDI musicians over to the IBM. It cost $495 at introduction, and played 8 FM voices. The quality of the FM was somewhat better than Adlib because it used a 4-operator FM chip (also from Yamaha) instead of Adlib's 2-operator chip, and had over 100 built-in instrument parameters. It also had a MIDI port. This board was, essentially, a Yahama FB-01 on a card. Trivia: You could put two of these boards in your PC at the same time to get a total of 16 simultaneous voices.

1988

Creative Music Systems (the name they had before they changed it to Creative Labs) came out with the Game Blaster around this time, and it offered 12 channels, with each channel producing either a single sine wave of a given frequency and magnitude (in stereo), or noise. The sound quality was obviously worse than the Adlib--the board simply couldn't do much of anything. You can still purchase CMS chips to put inside your Sound Blaster 1.x and 2.0, but it's really not worth it. The only game I know of that supported the CMS Game Blaster with decent music was was Times of Lore by Origin.

Digitized sound! Around this time, game companies had finally started to use digitized sound for music. (It had been used on the PC for sound effects as early as 1983, in Castle Wolfenstein/Beyond Castle Wolfenstein, and in 1987, in the PC version of Dark Castle.) While I had speculated that you could record bits of music and then rearrange them cleverly, a French game company called Loriciels beat me to it, with the excellent games Mach 3 (1987) and Space Racer (1988) (they also did a then-popular Pong/Breakout/Arkanoid clone called PopCorn). These games had a really cool (for a PC at the time) musical intro at the beginning, which was pieced together from small sound snippets that were arranged on the fly to form a longer piece of music. (You can think of this as a .MOD file with only one channel and all instruments/samples played at C#3.) And it played through the PC speaker! Coming from a PC that had a simple tone generator as a sound device, this just blew me away. Crazy Cars by Titus also had a snippet of digitized sound at the beginning, but this was just a 64K sample that looped once. The same went for Wizball by Mindscape. (Wizball was a fabulous game, IMHO.) Finally, games like Bop'n'Wrestle used it for the counts and body-slamming noises.

1989

The Sound Blaster hits the scene, and game companies start supporting it. It's essentially an Adlib clone, but it has the ability to record and playback digitized sound, allowing for speech and decent sound effects. (This information was essentially provided for people who don't know what a Sound Blaster is. I probably shouldn't have even written this paragraph, since the (in)famous Sound Blaster doesn't need mentioning, but I've done it already, so... whatever. :-)

1990

True Mixing. While many remember TrakBlaster being the first program to play MOD files on the PC, a couple of people were mixing before then, most notably PSI / Future Crew. He had created and marketed Scream Tracker in 1990 as shareware, and it could mix and play up to four voices in real-time on an 8MHz (or faster) computer. The early versions of Scream Tracker supported a mode of operation similar to the old SoundTracker on Amiga--you could save the song data and instruments separately. This allowed you to compose over 20 songs and fit them on the same disk, because you use the same set of instruments with each song. EGA Megademo / SpacePigs did this--they had four different songs that used the same instruments, so the whole thing fit onto a single 360K disk.

Take the PC Sound Standard Taste Test

All of the above is interesting, I agree, but it's "falling on deaf ears" (bwaahaa! I'm so damn funny!) if you can't get an idea for what the darn things sounded like. How was the Tandy/PCjr sound chip better than the PC speaker? Why did game manufacturers go through the trouble to provide sound support for the Roland MT-32? And just how good did PC speaker multi-voice attempts really sound?

Take the PC Sound Standard Taste Test! The following sound clips are not only examples of what each device/technique sounded like, but each category is specifically constructed from the same piece of music. This allows for easy comparison.

The Best Sound You've Never Heard

In the middle to late 1980's, software developers discovered that you could play digitized sound through the PC speaker without additional hardware by utilizing Pulse-Width-Modulation, which toggles the PC's internal speaker faster than it can physically move to simulate positions between fully on and fully off. (How's that for a clever hack?) Since you could get reasonable output of 6-bit digitized sound this way, some game developers decided to make use of it, either for sound effects (which was common) or for title music (which was rare). Some of these title "soundtracks" were original pieces of music composed specifically for the game, and many were extremely good for their time!

I've always been a fan of digitized title music in PC games. When all you've heard in your games are simple "beeps and boops" for a couple of years, the first time you hear real sound through your speaker, it floors you. It certainly left an impression on me, anyway. What also contributed to that impression was the uncommon quality of the music--it was usually well composed or thought out, which was not something you expected for a 4-color CGA game running on a 4.77MHz 8088 in 1987.

They were also cool from a technical standpoint: The programmer had successfully figured out how to output digitized sound through a device that was not designed to support such an action. How they discovered the technique of toggling the PC's internal speaker faster than it can physically move is beyond me. (Who was the first person to think of trying that?) Plus, the musicians themselves pulled some clever tricks, such as dynamic compression (in the sense of making quiet sounds louder, not the "pkzip" variety) and repeating sections, to not only make the title music better audible through the speaker, but also to make it longer. Space Racer is an excellent example of this: the actual audio data is 128K, but the entire title music (if saved into the same format) is much larger than that due to sections of music that can be "reused".

Sadly, most digitized title music of this nature has never been heard, since: 1. In most cases the game came out only for the PC before there was support for more conventional digitized output, such as the Sound Blaster or Covox Speech Thing, and 2. Most clone PCs had crappy psezio-electric "tweeter" speakers that resulted in barely audible or inaudible sound, instead of the large honkin' 3" speaker of the original PC. (The only exception to this is probably most Access titles, since they spent a lot of time creating and enhancing RealSound, their system of playing digitized sound on the PC, and eventually offered playback through the Sound Blaster and other devices.)

This section of oldskool.org is here to remedy that situation. In the following table, you'll find Sun/NeXT .au (u-law) sound clips of the title music of several games made during 1987-1989, which you can easily listen to right from your web browser through your sound device. While the sound is certainly not as polished as today's modern game music, the musical ideas and composition are certainly comparable. (Not to mention that, geez, this came out of your PC's speaker without additional hardware! Come on, isn't that cool as hell?)

Remember, these are not "mods" or "tracked music"; these are cleverly arranged, sometimes-repeating, single-channel sound samples. Enjoy!

Technical Notes

The Sun/NeXT .au format was chosen for these clips since, at the default 8000Hz sample rate, it reproduces all of the technical quality of the original sound, while at the same time being very cross-platform, easy to play, and has very low CPU requirements. RealAudio and MPEG Layer 3 were specifically not considered because their benefits (high audio clarity) were outweighted by their disadvantages (RealAudio is not as cross-platform as I would like, and MPEG Layer 3 has heavy CPU requirements during decompression). Besides, their additional clarity would not have represented the sound better anyway.

These clips were constructed from the original sound data wherever possible, which means that I took the raw PCM data files, cut and pasted them to match the output, then saved them into .au files. When that wasn't possible, they were recorded directly from the hardware through the use of 1/8th inch plug (sound card) to "alligator clips" (speaker) cable. While these are easy to make (Mean Streets even provides instructions in the README file), I just went out and bought one to save the trouble. The cable I purchased was less than four bucks from Radio Shack, Cat. No. 42-2421, description "Shielded 6-FT. (1.8m) AUDIO CABLE; For Speciality/replacement audio connections; 1/8" mini plug to alligator clips", and it works quite well; I didn't need to splice a capacitor into the cable because I was recording with my sound card, which was able to raise the input level. I attached the ground clip (the one attached to the copper shielding) to the case and the other clip to one of the speaker terminals. It works great, IMO.

The only drawback to recording sound from old PCs via this method is that you usually have to do so with the cover off, and old PCs were hardly FCC Class A certified, let alone Class B. The result? If you're lucky, you get only a small amount of noise during the silence portions that can be removed with a simple noise gate. If you're unlucky, however, you'll get a noise or static that's audible during the non-silent portions. If you're cursed and/or improperly grounded, you'll get a nasty 60Hz line hum from the wall AC. Post-processing with a good set of software filters is the only way to remove noise/distortions like that, but if you don't have a good set of software filters, you can always simply resample your original digitized recording down to about 8000Hz--this will get rid of mostly everything except the 60Hz AC hum.

Similar Pages

The following is a list of pages similar to this one:

MegaMan_X's Soundcard Hall of Fame has some nice pictures of sound cards, as well as some additional information.