Create your own
Lesson illustration

Evolving Textures in Streets of Rage 2: Koshiro's Real-time Modulation

Hello! Welcome to the final lesson in our case study on Yuzo Koshiro.

In our last session, we analyzed the significant evolution in sound design and composition from Streets of Rage to Streets of Rage 2. We noted the shift towards harder techno influences and identified how improved hardware access (more memory for samples) and more aggressive FM patch design contributed to this change. We also touched upon the idea of emulating analog filter sweeps by dynamically changing FM parameters—a technique that serves as a perfect entry point for today's topic.

This lesson will focus on a single, advanced concept that is arguably the pinnacle of Koshiro's technical artistry on the Genesis. We will address the learning outcome: Analyze a specific example of Koshiro's use of real-time parameter modulation to create evolving textures in 'Streets of Rage 2'. We will define this technique, look at a modern demonstration, and then deconstruct how Koshiro applied it in a far more complex way, effectively turning the YM2612 into a semi-generative instrument.

1. The Principle: Making Sounds Evolve in Real-Time

At its core, "real-time parameter modulation" means changing the parameters of a synthesizer patch while a note is playing to alter its timbre. This goes beyond the static, pre-set change of a standard ADSR envelope. It's about creating a sound that feels alive, dynamic, and unpredictable.

The most common way to do this is with a Low-Frequency Oscillator (LFO), which can be programmed to rhythmically or smoothly alter parameters like volume (tremolo), pitch (vibrato), or, most interestingly for FM synthesis, the amplitude of a modulating operator to change the harmonic content.

To see this principle in a clear, isolated context, let's look at a modern example.

How to actually use FM synthesis — A practical guide

This video from the channel cooksonia provides a practical guide to FM synthesis. We will watch a specific segment where the creator builds a dub chord, using an LFO to create a 'continually evolving sound'. This is the fundamental concept we will build upon.

Please watch the segment from 00:47:27 to 00:52:01. Pay special attention to the explanation starting around 00:48:40. The creator uses a slow LFO to modulate the volume (level) of a high-frequency operator, causing the bright, harmonic part of the chord to fade in and out unpredictably. This creates timbral movement over time.

This video demonstrates the effect perfectly: a simple LFO applied to an operator's level makes the sound dynamic. Now, imagine doing this not with a simple, repeating LFO, but with a custom-programmed, complex sequence of changes. This is where Koshiro's genius lay.

2. Koshiro's Secret Weapon: The "Programmer's LFO"

The YM2612 chip has a built-in hardware LFO, but it is quite basic, offering a limited frequency range and only a sine waveform for modulation. While useful for simple vibrato and tremolo, it could not create the complex, sometimes chaotic textures heard in Koshiro's work.

Koshiro's key innovation was to bypass the hardware LFO and create his own modulation sources in software. Using the Genesis's dedicated sound CPU (a Z80), he wrote custom code that could directly and rapidly update the YM2612's parameter registers during note playback. This "programmer's LFO" could have any shape, speed, or complexity he desired. He wasn't just composing notes; he was programming the synthesis engine itself as part of the composition.

We have direct evidence of this programming-centric approach from Koshiro himself.

Yuzo Koshiro – 2001 Composer Interview

This 2001 interview from shmuplations.com is one of the most insightful sources on Koshiro's methods. He explicitly discusses his custom programming environment and his experiments with algorithmic music generation.

Please read the two sections starting with 'Has your musical workflow changed...' and 'During Bare Knuckle 3, I actually wrote a "self-composing" program...'. Focus on these key points: His custom version of Music Macro Language (MML) allowed him to 'deeply customize every aspect' and 're-create' hardware functionality in software. His claim that 'with some programming tricks I can make it sound like 6 voices'—a direct statement about using software to transcend hardware limitations. His description of the 'self-composing' program for Streets of Rage 3, which used 'random numbers and mathematical formulas' to generate musical phrases. This reveals his interest in algorithmic and generative techniques.

This interview confirms that Koshiro was not just a musician using a tool; he was a programmer modifying the tool itself. The "self-composing" program for SoR3 is the ultimate expression of this, but the foundations for it were clearly laid in SoR2.

3. Case Study: The Evolving Textures of Streets of Rage 2

Now, let's connect this technique to a specific sound. While we don't have an isolated track in our resources, we can analyze the technique used for one of the most iconic sounds in Streets of Rage 2. If you listen to the track "Alien Power" (from Stage 6), you will hear a high-pitched, metallic, arpeggiated synth line. The timbre of this sound is not static; it shifts and morphs constantly, creating a feeling of alien, unpredictable energy.

How was this achieved? Let's build a strong hypothesis based on our evidence.

  1. The Foundation: The sound is a standard 4-operator FM patch. The metallic quality comes from using operators with non-integer frequency ratios, creating an inharmonic spectrum.
  2. The Modulation: The "evolving" texture is the result of real-time parameter modulation. Specifically, Koshiro's Z80 driver is likely targeting the Total Level (TL) parameter of one or more of the modulator operators.
  3. The "Programmer's LFO" in Action: Instead of a smooth, predictable sine or triangle wave LFO, Koshiro's code is likely feeding a pseudo-random or pre-defined sequence of values to the TL register. Each new value changes the modulation index, which in turn alters the number and amplitude of the harmonic/inharmonic sidebands.
    • A high TL value (less attenuation) means a lower operator output level, resulting in a simpler, purer tone.
    • A low TL value (more attenuation) means a higher operator output level, resulting in a complex, bright, and noisy tone.

By stepping through a sequence of these TL values very quickly—perhaps synchronized to the 16th notes of the arpeggio—Koshiro creates a sound that glitters and transforms with every note. This is the "self-composing" concept from the interview applied to timbre instead of melody. He defined a set of rules and data, and the program "performed" the timbral changes automatically.

Another advanced technique Koshiro likely used, especially for the vocal-like sounds in Streets of Rage 3, is the YM2612's "CSM Mode" or "Illegal Mode." This involves rapidly re-triggering operators to create formant-like effects. You can read about this in the SEGA FM DRIVE TECH MANUAL (resource e8e71, part 3). While he may have experimented with it in SoR2, the technique described above—a custom software LFO modulating operator levels—is a more direct explanation for many of the game's signature evolving textures.

Conclusion

Yuzo Koshiro's use of real-time parameter modulation represents a fusion of high-level programming and sophisticated sound design. He didn't just compose music for the Genesis; he programmed new sonic capabilities into it.

Key Takeaways:

  • Real-time parameter modulation involves changing synthesis parameters (like operator level, frequency, or feedback) while a note is playing to create dynamic timbres.
  • While the YM2612 had a basic hardware LFO, Koshiro's innovation was the "Programmer's LFO"—custom Z80 code that could manipulate any parameter with arbitrary complexity.
  • His own interviews confirm his deep programming expertise and interest in using algorithmic and generative methods to create music.
  • A specific example is the evolving arpeggio in SoR2's "Alien Power," likely created by a software routine rapidly changing the Total Level (TL) of modulator operators, thus altering the sound's harmonic complexity in real-time.

This deep, technical approach stands in fascinating contrast to the methods of other great composers of the era, which we will explore next.

Next Lesson Preview:

This concludes our deep dive into the work of Yuzo Koshiro. In the next module, we will turn our attention to another giant of Sega music: Masato Nakamura. We will explore how his background as a professional bassist and hit-making composer for the J-pop band Dreams Come True led to a completely different, yet equally iconic, sound for the Sonic the Hedgehog series. We will shift our focus from technical programming to the art of pop songwriting and melodic craftsmanship on the Genesis.

Can't find a good explanation? Sign up and we'll make it for you

Sign up