What Is FM Synthesis and How Does It Work?

Frequency Modulation (FM) synthesis generates sound by using the frequency of one waveform (the modulator) to modulate the frequency of another waveform (the carrier). Unlike subtractive synthesis, which filters harmonically rich waveforms, FM synthesis creates complex harmonic structures by varying the modulation index and the ratio between the carrier and modulator frequencies. This technique was famously commercialized by Yamaha in the 1980s with the DX7, which defined the sound of pop, jazz, and film scores for decades. Understanding FM is essential for any modern sound designer or music producer because it offers a unique ability to produce metallic, percussive, and evolving timbres that are difficult to achieve with other methods. For a thorough technical overview, refer to Sound On Sound’s definitive guide to FM synthesis.

The Core Components: Carrier, Modulator, and Modulation Index

Every FM patch consists of at least two oscillators. The carrier oscillator outputs the audible sound we hear. The modulator oscillator provides a control voltage that changes the carrier’s frequency at the rate of its own frequency. The amount of this frequency change is called the modulation index. A higher modulation index introduces more sideband frequencies, which add brightness, grit, and harmonic complexity. A lower index keeps the sound closer to a pure sine wave. The ratio between the carrier and modulator frequencies determines the harmonic relationship. Ratios like 1:1 produce overtones similar to a sawtooth wave, while non-integer ratios (e.g., 1:1.4) produce inharmonic partials — perfect for bell-like sounds.

Why Ratios Matter for Bell and Plucked Sounds

Bell sounds rely on inharmonic partials — frequencies that are not integer multiples of the fundamental. A classic bell patch might use a carrier-modulator ratio of 1:1.43, 1:2.7, or 1:4.2. These ratios create partials that sound metallic and resonant, mimicking the vibrational modes of real metal. For plucked string sounds, you want a slightly harmonic or quasi-harmonic structure. Ratios such as 1:2, 1:3, or 1:4 produce bright, twangy attacks that decay naturally. A useful trick is to start with a simple 1:2 ratio (carrier = 220 Hz, modulator = 440 Hz) and then vary the modulator frequency by a few cents to introduce subtle inharmonicity — this gives a more realistic string sound. Detailed ratio charts can be found in the Yamaha Synth Academy’s FM tutorial.

Setting Up Your Environment

Before diving into programming, ensure you have an FM synthesis tool. Many modern DAWs include built-in FM synths (e.g., Ableton Live’s Operator, Logic Pro’s EFM1, FL Studio’s Sytrus). Free options like Dexed (a Yamaha DX7 emulator) or VCV Rack’s FM modules are also excellent. Initialize a patch: set both carrier and modulator to a pure sine wave, no filtering, and unity gain. Start with a single modulator feeding a single carrier (a two-operator configuration). Later we will add a second modulator or feedback to refine the sound. Keep an A/B reference — compare your sound to a real bell recording or a plucked guitar sample. Use an oscilloscope or spectrum analyzer to visualize the partials. This step-by-step approach prevents overwhelming complexity.

Step 1: Bell Sound — Choosing the Ratio and Modulation Index

For a classic bell, set the carrier frequency to a pitch you want the bell to ring at, say C5 (523.25 Hz). Set the modulator’s frequency to a ratio of approximately 1:4.0 or 1:3.76. A 1:4 ratio gives strong sidebands at integer intervals, but a slight detune (e.g., 3.95) yields more metallic richness. Increase the modulation index gradually. Start with a value of 2.0 and listen. You should hear a bright, clangorous sound. Adjust the index between 1.5 and 5.0 while playing a note. Higher indices produce more upper partials and a harsher tone — ideal for large church bells. Lower indices (around 1.0) create softer, more mallet-like bell tones. Important: The modulation index is often controlled by an envelope. For bell sounds, the index should start high and decay quickly, mirroring the natural decay of a struck bell. Set the modulator’s amplitude envelope to a fast attack (0–5 ms) and a decay that matches the desired bell length (e.g., 2–4 seconds). The carrier’s amplitude envelope can be similar but slightly longer to allow the harmonics to fade naturally.

Fine-Tuning with Additional Operators

Real bells have more than two resonant modes. Add a second modulator with a different ratio (e.g., 1:2.1 or 1:5.6) and a lower modulation index (0.5–1.5). Route both modulators to the same carrier. This adds extra partials that fill out the bell’s body. Avoid making the modulation index too high on the second operator, or the sound becomes noisy. Listen to classic DX7 bell presets like “FullTines” or “Marbell” for inspiration. Many free patch libraries are available online.

Step 2: Plucked String Sound — Envelope Shapes and Ratio Selection

Plucked string sounds (acoustic guitar, harpsichord, koto) require a sharp attack followed by a rapid decay to a lower sustain level. The key is to use a modulator-to-carrier ratio that mimics the harmonic series but with a slight offset for realism. Start with a ratio of 1:2.0 for a bright, twangy tone. A ratio of 1:3.0 produces a thinner, more nasal sound reminiscent of a harpsichord. Set the initial modulation index high (e.g., 4.0) and use an envelope to drop it quickly to 0.5 over 100–300 ms. This simulates the string being stretched and then damping. The carrier’s amplitude envelope should have a fast attack (1–5 ms), a decay that matches the index envelope (around 200 ms), a sustain level of 0.2–0.4, and a short release. The result is a percussive attack that fades into a softer, more fundamental tone.

Adding Realism with Feedback and Detuning

FM synthesis for plucked sounds often benefits from operator feedback. Route the carrier’s output back into its own frequency input. A small amount of positive feedback (0.1–0.3) adds a slight grating, quasi-physical sensation — like the string scraping against the fret. Experiment with negative feedback for a more mellow effect. Another technique: detune a second carrier (or an additional operator) by 5–10 cents. This creates chorus-like movement that mimics the beating of real strings. Use a low modulation index on this second voice to avoid muddying the attack. Set its ratio identical to the primary carrier and detune it slightly. Many hardware FM synths call this “unison” or “stacked voice.”

Step 3: Envelope Shaping for Natural Decay

The envelope is arguably the most critical part of FM sound design. For both bells and plucked strings, the modulator and carrier envelopes must work in tandem. Use four-stage envelopes (ADSR) with accurate timing. A general rule: the modulator envelope’s decay time should be half the carrier’s decay time for plucked sounds. For bells, the modulator’s decay can be shorter (since high harmonics die out faster) or even the same length if you want a sustained ringing. Set attack times to 0 for percussive sounds. For softer mallets, attack times up to 10 ms can soften the blow. Allow enough release time (500 ms–1 s) so the sound doesn’t clip unnaturally. Use a negative envelope polarity on the modulation index if your synth supports it — this can create an inverted effect where the sound brightens over time, useful for “crackling” bells.

Step 4: Advanced Modulation — Velocity Sensitivity and LFO

Expressiveness comes from dynamic control. Assign the note velocity to modulate the modulation index (or overall output level). Higher velocities should increase the index, producing brighter, more aggressive sounds. For a harpsichord-like pluck, map velocity to both the carrier level and the modulator’s envelope depth. Add a slow LFO (0.1–0.5 Hz) to slightly modulate the carrier or modulator frequency. Depth should be very small (1–5 cents). This adds a subtle shimmer or wavering pitch, making the sound feel alive. Avoid fast LFOs — they turn into vibrato, which is not typical for percussive strikes.

Step 5: Processing and Effects Chain

Raw FM patches often sound thin or harsh. Apply effects to simulate an acoustic environment. Start with a convolution reverb with an impulse response of a small hall or plate. Reverb tail length of 1–2 seconds for plucked sounds, 3–5 seconds for bells. Use a low-pass filter before the reverb to tame high-frequency harshness; set cutoff around 8–12 kHz. A touch of stereo delay (80–120 ms) with feedback of 0.1–0.3 can widen the soundscape. For plucked sounds, add a subtle compressor with a fast attack (10 ms) and medium ratio (3:1) to even out dynamics. Avoid over-processing — FM’s charm is its clarity. For an in-depth look at signal chains for FM sounds, see Attack Magazine’s FM sound design techniques.

Troubleshooting Common Issues

  • Harsh or piercing tone: Lower the modulation index or reduce the modulator’s ratio to a less inharmonic value. Alternatively, use a low-pass filter with a gentle slope.
  • Muddy attack: The modulator envelope’s attack may be too slow. Set it to 0 or 1 ms. Also check that the carrier’s envelope isn’t too long on the attack stage.
  • No percussive attack: Ensure the modulator’s initial index is high enough. Many synths have a separate “initial level” parameter for the envelope — set it to maximum.
  • Unstable pitch: Some FM synths allow the modulator to be slightly out of tune. Use a fixed frequency mode instead of ratio mode if pitch drift occurs. Also check for unintended LFO modulation.
  • Aliasing artifacts: If you hear high-frequency whine, reduce the modulation index or use an anti-aliasing option in your synth. Some software synthesizers oversample internally — enable that.

Practical Examples and Patch Construction

Example 1: Small Church Bell

  1. Carrier sine wave, frequency 261.63 Hz (C4).
  2. Modulator 1: ratio 1:3.76, waveform sine, initial modulation index 3.0. Envelope: attack 0 ms, decay 3 s, sustain 0, release 500 ms.
  3. Modulator 2: ratio 1:2.1, index 1.0, envelope: attack 0 ms, decay 1.5 s, sustain 0, release 300 ms.
  4. Route both modulators to the carrier (additive FM).
  5. Carrier envelope: attack 1 ms, decay 4 s, sustain 0, release 1 s.
  6. Add reverb: small hall, 2.5 s decay, wet 30%.

Example 2: Acoustic Guitar Pluck

  1. Carrier sine wave, frequency 110 Hz (A2) for low string, 440 Hz for high.
  2. Modulator: ratio 1:2.005 (slightly over), index initial 4.5. Envelope: attack 0 ms, decay 250 ms, sustain 0.3, release 400 ms.
  3. Enable carrier feedback at 0.15.
  4. Carrier envelope: attack 2 ms, decay 800 ms, sustain 0.2, release 500 ms.
  5. Detune a second carrier by 8 cents, same settings, mix 50%.
  6. Add a slight EQ boost at 2.5 kHz (bell curve, +3 dB) for presence.
  7. Compressor: attack 5 ms, release 150 ms, ratio 4:1, threshold -12 dB.

Further Exploration and Resources

FM synthesis rewards experimentation. Try different waveforms for modulators (saw, square, triangle) — they add extra harmonics, but can make the sound chaotic. Use a sine wave for clean results. Study the classic DX7 patches: many are available as .syx files that can be loaded into Dexed. Patch Library offers thousands of user-submitted FM presets to analyze and modify. For a comprehensive textbook approach, read FM Theory and Applications by John Chowning, the inventor of FM synthesis. Also, check out Ableton’s interactive music theory site for a primer on harmonics related to FM. Remember, the most important skill is critical listening: compare your sounds to acoustic references and trust your ears. Over time, you will develop an intuitive sense for which ratios, indices, and envelopes produce the sounds you imagine.

Final Thoughts on Programming Bell and Plucked Sounds

FM synthesis is both mathematical and musical. The steps above give you a reliable framework, but the real magic happens when you deviate intentionally. Start with the ratios and indices suggested, then tweak each parameter in small increments. Use a real audio spectrum tool (e.g., Voxengo SPAN) to see the partials you’re generating. Eventually, you’ll be able to dial in a bell or plucked sound in under a minute. Practice on different carrier frequencies — low bells and high bells require different ratios and index envelopes. Plucked sounds on bass frequencies need slower decays and less feedback to avoid flabbiness. With consistent experimentation, you will master the classic bell and plucked string timbres that made FM synthesis a cornerstone of electronic music.