Retro Game Audio

The SNES has a lesser-known audio feature called pitch modulation. Similar to 2-operator FM synthesis, the values from one sound channel can alter the output of another. This is useful for distorting the sound, which adds some versatility to the audio beyond the “bare bones” sample playback. 

The SNES was capable of adding echo effects to its sound channels.

Example #1: Mega Man X Music

In this first example, I demonstrate what the SNES’ built-in echo & reverb effects contribute to the overall sound, by toggling the effect every four measures. On some speakers, particularly mobile devices, it may be hard to hear the difference. But on decent speakers and headphones, the difference is very clear.

Example #2: Chrono Trigger Title Screen

The Immortal is the only game in the entire Famicom/ NES library to use a looped sample to generate a sustaining, melodic waveform. I say this with some degree of confidence because, even though I can’t say that I’ve probed every track of music from every single Famicom and NES game, it’s been years since I’ve made this observation and no other examples have been found by the community! Pretty cool stuff! Leave it to Rob Hubbard to do something cool and unique with the audio.

This is something I’ve posted about before, though I believe I erroneously referred to the sample as being only 1 byte in size. In retrospect, how would that be possible?! It’s because the .dmc ripping tool I used at the time didn’t locate the proper sample, so instead I was looking at the wrong thing and didn’t realize it. Nonetheless, it still is a very small looped waveform.

I’ve started a series of videos on twitter and youtube with the hashtag #VGMTrivia. The idea is to share bite-sized nuggets of random-but-cool information on classic video game music. Hope you enjoy!

https://twitter.com/hashtag/VGMTrivia?src=hash

Battletoads is known for using 7-bit PCM, which was a less common method for sample playback than the more typical 1-bit DPCM found in NES games.

Raw PCM is more demanding of resources than DPCM, usually freezing gameplay since the NES can't handle performing other actions simultaneously. This is why PCM a drum beat shows up in the "pause" music of Battletoads, since it's a convenient time for it.

But for the introduction of Battletoads, they decided to try cram in a PCM bass drum and snare wherever they could. Using rainwarrior's lua script to display when PCM is active (the display in the top left), it's easy to see and hear when the PCM has to drop out because of the visuals that are too complex to run simultaneously.

Episode 21 of Retro Game Audio is live.  Steve and I were joined by Andrew Reitano to discuss high quality sample playback on the NES.

Something pretty well-known about the triangle channel on the NES is that it doesn’t offer proper volume control (it can be messed with, slightly, but that’s another story). But I never really considered why this was the case.

Thankfully, this was recently explained to me by hunretrogeek, and the answer is surprisingly simple: It’s already using up the allotted volume space to make the waveform itself.

Consider the square/pulse waveforms on the NES. There’s not really any verticality to them. They’re either in the up position, or the down position, never anywhere inbetween. They were also given a 4-bit space for volume control, resulting in 16 possible volumes (0 = silent, 1 = softest, 15 = loudest).

The triangle wave? It was also given a 4-bit space to work with. But it’s already eaten it up in the process of making the triangle shape:

This is a mix a friend of mine made a couple years ago, originally uploaded to bandcamp: https://kirbypufocia.bandcamp.com/album/cviii-no-densetsu-or-whatever

Castlevania III was originally released in Japan as Akumajou Densetsu (悪魔城伝説), and it used the VRC6 for sound expansion. This gave it three extra channels of audio, which primarily included voices not found on the NES. Though many people prefer the original, fuller version of the soundtrack, each version has its pros and cons. This mix seeks to implement the best of both worlds by carefully mixing both versions together, and re-balancing the mix. 

This video is a chronological journey through Konami's Famicom and NES soundtracks, playing one song from every game. It also includes text commentary on the sound design and history.

We’re back! In pog podcast form! 

After a long hiatus from this blog, and chip music stuff in general, I’d been considering returning with a renewed interested in podcast-form. However, I hadn’t really given it serious consideration until after I met Steve Lakawicz at a show recently in Philly.

Steve has a blog that also took a break, Classical Gaming, which has a lot of great content with a similar theme to the sort of stuff I posted here and on explod-blog. After talking about video game music with him for a while, I realized I’d found a great potential co-host. He also makes excellent music under the name Ap0c, which I highly recommend checking out. 

Our first episode will be a close up look a specific NES soundtrack, doing a track-by-track analysis of the sound design, and it should likely be up on Thursday:

Can you explain what it means, exactly, for music to be "8-bit?" What significance does the number of bits have for waveforms, channels, sample size/quality, volume, speed, etc? In other words, the number of bits limits _____ to _____?

This is a great question, I'll try to answer as best as I can.

1. What is 8-bit music? 16-bit music?

In general terms, I'd argue that these titles are meaningless by themselves. A lot of people associate "8-bit" with things that sound like the NES, but the problem with this is that the NES doesn't primarily sound the way that it does because it's 8-bit, it sounds the way it does because of the particular kinds of sounds it uses.

It's not that the limitations don't matter at all, the NES would have had different audio if the system belonged to a different 'bit generation', but here's an example of why these titles don't mean much on their own:

Let's say you have a 16-bit wave file of CD music and you convert it to 8-bit. Is the end result something that sounds like the NES? Not at all.

This is why specificity matters. Classic video game systems sound the way they do because of the approach to building or creating sounds. Is the system using sound synthesis? If it is, is it perhaps using subtractive synthesis, FM synthesis, or wavetable synthesis? Or is the system playing back samples, or some combination of different kinds of synthesis and samples?

Once you determine these distinctions, this tells you why a given system sounds the way it does.

2. Another way of putting it, is that systems can sound similar or dissimilar regardless of which bit generation they belong to.

Another way to illustrate my first point would be to share a few examples. Consider Amiga music:

This is 8-bit music, but it doesn't sound anything like the NES. This is because it is music that's sample-based, making it more comparable to the SNES (SNES uses 8 channels of 16-bit sample playback, Amiga uses 4 channels of 8-bit sample playback).

Another good example is the Virtual Boy.

A while back, I was trying to find out more about the audio behind the Virtual Boy, and was disappointed to find that most sites, including wikipedia, had nothing more to say than "16 bit stereo audio".

That doesn't tell me practically anything. The SNES is also 16-bit, as is the Sega Genesis, which all sound different from each other (it turns out that the Virtual Boy uses 6 channels of primarily wavetable synthesis).

Last but not least is the Game Boy. It is the closest thing there is to the NES in terms of audio, but it's 4-bit, not 8-bit.

3. But the bits do matter...

Before I give the false impression that the 'bit limitations' of a system aren't incredibly important to the sound capabilities (too late?), I'll need to clarify that the amount of bits available that can be manipulated are the be-all end-all of what you can do.

Then why spend so much time downplaying the titles of "8-bit" and "16-bit"? It's because these titles can be misleading, as they often don't tell you about the specific limitations of different parameters.

THE NES / Famicom

Let's consider volume on the NES. If we were to assume that the NES was 8-bit everything, then we'd expect there to be 256 different possible volume settings for a given voice. But that's not the case, the channel volume is determined by a 4-bit value, allowing 16 possible volumes (0-15) for the pulse waves and noise channels.

What does this mean in musical terms? Lower bits means less resolution. Let's say you wanted to have a note fade out. If you start with a high or medium volume, you have room to descend and your ear will hear a smooth-enough fade out. But if you started with a quiet note and wanted to do a long fade out, there are no values in-between what's available. If you go from volume 3, to 2, to 1, and then 0, and you don't do this quickly, you won't hear a smooth fade out at all. The volume will distinctly "jump" between those values, with no means for a smoother fade (unless maybe you try to cover it up with other sounds at the same time).

There's also the sample channel on the NES. Those samples aren't 8-bit either, they're actually just 1-bit/ 7-bit... (more on that in a future post)

A quick search on the NESdev page that details the NES APU will yield 5 matches for 4-bit and 0 matches for 8-bit. Go figure.

Wavetable Synthesis

Some systems use wavetable synthesis, which in layman's terms is a sort of free space to draw whatever waveform you wanted. In software, it often looks like a bar graph.

-The Game Boy allows for a table of 32 4-bit samples to create a waveform (in one channel).

-The Virtual Boy allows for a table of 32 6-bit samples to create a waveform (in five channels).

-The Famicom Disk System add-on adds one additional channel to the Famicom that uses 64 6-bit samples.

Even though these are all examples of wavetable synthesis, the bit-limitations create sounds that are distinct from each other. In a future post, I may compare the wavetable synthesis of these three systems more closely. The basic summary is that the lowest resolution of the three is found on the gameboy, which has the "dirtiest" sound. By comparison, the Famicom Disk System can create some very smooth sounds.

- - -

I hope this answers your question. The number of bits limits your resolution, which affects the qualities and potential for all sorts of parameters.

The same basic trick applies to the snare drum, but it usually involves higher pitches, sometimes with a shorter duration.

If newer readers haven't seen it yet, you can watch the video in this post for a demonstration of how the triangle wave is used for percussion.

Comparing kick vs. snare triangle effects in several games, I've found that the snare sound was actually pretty similar in all of them. The "sweet spot", at least for these examples, seems to be starting somewhere around G3-A#3, and then pitching down only within that octave, or going down somewhat into the octave beneath but not very far.

^Asterix Triangle Snare - A#3, G3, E3, F#2

^Hero Quest Triangle Snare - G#3, F#3

^Solstice Triangle Snare - G3, F#3

If you want to compare these snare drums to the bass drum sounds used in these games, I uploaded an album showing both:

http://imgur.com/a/hc5ms

Thank you! Solstice has an excellent soundtrack composed by Tim Follin. He is highly regarded for his "clever manipulation of limited sound hardware", as wikipedia puts it.

The big sound comes from a combination of techniques; we can use the title theme to demonstrate them.

At around 8 seconds in, we hear an explosion of colorful sounds. It's a drastic change from the introduction, as if it's mocking the simplistic sound.

1. Using the Triangle Wave to Augment Percussion

The triangle channel, which is most commonly used for basslines in NES music, is used to make the drums sound a lot bigger and heavier. This is achieved by adding quick downward pitch bends to create bass drum and snare drum sounds (that double up with the noise channel).

If you were to isolate the triangle channel, you would hear that the basslines are clearly interrupted by these percussive sounds. But when everything is put together, your ear is never really bothered by the fact the bass never hits at the exact same moment as a bass drum or snare drum. It's always a little behind or "in-between", and the end result is a full sound with big drums.

This previous post explains triangle percussion in more detail, with a video demonstration.

2. The Arpeggio Effect

Rapidly alternating notes are used to simulate chords in the NES' monophonic channels. To overcome the limitation of not being able to play a chord in a single channel, chords are broken up into their individual notes and changed very quickly.

Solstice doesn't use this effect timidly, instead it it opens up after the introduction with two different arpeggios stacked on top of each other. At other times, one of the pulse waves may be playing a melody while the other supports it with arpeggios behind it.

More information on this effect here.

3. Dynamics

Including voices that are quiet go a long way in making this track sound bigger. If everything was loud, you'd adjust your speakers accordingly and there'd be little room to go up.

Apologies to those who've had unanswered questions. I didn't intend to take such a long break from this blog. I hadn't logged into my youtube channel or tumblr for quite a while, but I'm happy to discover newer followers and questions being submitted in that time.

I'll try to answer what questions I can over the next couple days, and hope to start posting again somewhat regularly. Unrelated, here's a gif I made from Flashback: