Audio/game sound effects -how were they made?

Short answer is that it's not just square waves. That would only create one frequency note (plus its odd harmonics).

The key to more sophisticated audio effects is to use pulse-width modulation. This can create much more intricate waveforms especially after the normal filtering effects of the Apple speaker circuit. Depending on how complex the software is, you can expect to achieve multi-voice harmonies and acceptable speech synthesis with just a 1-bit toggling speaker citcuit!

Check this out.

You've got $C030 accesses and time in microseconds.  Machine language, cycle counting all to fiddle with the timing between the clicks of the speaker.

"1-bit" audio generation is not the limitation it seems. The highest fidelity systems like DSD use 1-bit sampling at multi-MHz rates.

More modestly, there are demos on home computers that play back 4-channel sampled tracks through the "1-bit" speaker channel.

And there are oddball art projects that create some quite interesting music by toggling a single pin on a microcontroller.

Fuji wrote:

So what's the secret behind sound effects in games such as  Lode runner, Gumball, Choplifter and Ms. Pacman (to name just a few)?

When I beat games back in the 80's, I then reverse-engineered as a second challenge.  So I learned how two of your listed examples worked by deciphering their machine code...

• Gumball simply played its music in truncated, staccato notes with silent gaps between them -- the program did its animation (and other computing) during the silent gaps between notes.  That's why Gumball always pauses when it plays the factory whistle -- the whistle is a continuous sound effect, without any silent gaps, so there's no animation until the sound finishes.  This is a very common method of generating sounds in Apple games.
• Ms. Pacman  uses a unique duet music library that was (AFAIK) only used in Atarisoft games.  Its machine code kept track of two counters for the two musical notes to be played at the same time, and used the carry bit to keep track of which direction the 'current' polarity of the Apple speaker.  Each time one of the notes called for generating an "edge" to the output waveform, it would compare the edge's direction the carry bit to decide whether to pulse the speaker and thereby generate a new edge.  If it encountered two adjacent edges in the same polarity, it simply merged them together into a single edge in order to avoid scrambling the signal.  The Atarisoft music generator was therefore just as loud as normal sound effects, unlike most multi-voice sound generators (eg: Electric Duet, Music Construction Set, etc.).

jeffmazur wrote:

Short answer is that it's not just square waves. That would only create one frequency note (plus its odd harmonics).

 

 

Thanks for the link to Paul Lutus' Electric Duet.  It brought back some great memories and reaquainteed me with Paul's great contributions.

Fuji wrote:

Did you manage to extract the sequences ("melodies") from those games?

 

No, I didn't think of extracting the melodies.  I enjoyed making it replay its own songs, but I only disassembled the subroutines that actually played the musical notes out of the speaker.  And I "borrowed" their ideas to make my own music and sound effect routines.

This YouTube clip from Moon Patrol illustrates several aspects of Apple sound fx.  The background melody is played in tiny staccato pulses.  As the buggy passes checkpoint Y, the program has to do more animation work, so the pulses of music become audibly sparser (and slower).  Then when the buggy reaches checkpoint Z, all animation stops in order to play a lively duet using the sound routine explained below.

Around 1984 I disassembled the two-voice music subroutine Atarisoft had used in Moon Patrol -- it was the same subroutine that Ms Pacman used in your "Ms. Pac-Man 1" sample.  Here's an illustration of how its algorithm works, as far as I remember...

Suppose Atarisoft's "duet" music routine was trying to play two musical pitches illustrated by the green square wave and the purple square wave.  The duet routine doesn't just blindly pulse the speaker for both square waves because it would create a chaotic jumble of extra edges, sounding nothing like either melody.  Instead, the routine pays attention to whether each edge is "rising" or "falling".  For instance, the first time the green wave rises, the routine pulses the speaker to make it rise too.  But then when the purple edge rises, the routine does not pulse the speaker because it has already generated a rising edge and doesn't need another rising edge.

Square waves.png

Fuji wrote:Regardless of having that data converted into a MIDI file or having to manually enter the information (note, duration, next note (or pause), duration etc.) it would be very useful for learning how they were constructed.