Synthesizing a metallic clicking noise

Question

Bottom line up front:

I would like to synthesize the metallic click of a relay (similar to old school turn signal noise). I can get a click (white noise->Short attack/release->bandpass) but it's missing the metallic part, any hints would be appreciated.

Full Story:

New to synthesis, I bought bitwig for its polygrid as a playground so that's my major exposure and how I'll frame what I've done so far.

After I got past simple stuff, building cross fading synths, etc. I listened to a song with a clicking noise and it reminded me of what I originally got exposed to synthesis for: I wanted to create a turn signal clicker

These come in many variations today (some vaguely melodic) but I mean the old school "relay" noise. I got the click from the song by putting white noise through a very short Attack-Release period and a bandpass. But for a turn signal it would be missing that characteristic metallic-ness of a relay reed slamming against a contact.

Any ideas on how to get that effect would be greatly appreciated.

Answer 1

In addition to Tetsujin's comment about ring modulation getting you a metallic sound, frequency modulation is another great way. Specifically, modulating one oscillator's frequency with a second, audio-rate oscillator.

Both modulations create sidebands that are not necessarily harmonic with the fundamental/original frequency, which is a strong characteristic of struck metallic sounds.

A great first place to look for help in synthesis of sounds is Sound on Sound's Synth Secrets series of articles. Here's the relevant article on metallic synthesis (cymbals): https://www.soundonsound.com/techniques/synthesizing-realistic-cymbals

Quoting liberally so this is not a link-only answer:

We know from last month's studies that the cymbal's sound is a dense fog of [inharmonic] partials, and that it is not dominated by any particular modes. The easiest way to produce these is by using FM synthesis.

(later)

Figure 2 shows the oscillators I have chosen for the patch. As you can see, I have directed the output from the modulator ('OscA1') to both the pitch input and the FM input of the carrier ('OscA2'), and set the amplitude of the modulation to maximum in both cases. Using both the Nord's linear and logarithmic modulation inputs ensures that the resulting spectrum is suitably complex.

Note that for most synthesizers (real and virtual), an oscillator's frequency control input is usually logarithmic, so it can easily follow the pitch ratios that make sense for musical notes. But an FM control input is usually linear, which is more appropriate for modulations of this kind. That's what the last sentence of the above quote refers to.

(later)

Figure 3 shows the architecture I have used to generate the initial 'ping' of the cymbal. This passes the carrier's output through a 24dB-per-octave band-pass filter. I have set the base frequency of this to be approximately 1kHz, and then applied a fast Attack-Decay envelope that sweeps the pass-band down to 1kHz over the course of just one-fifth of a second. The attack is as near-instantaneous as it is possible for a Nord envelope to be.

(later)

So, how does it sound? To be honest, the result is not as realistic as I had hoped, largely because the initial FM spectrum is not quite close enough to the real thing. Instead, this patch is not unlike that obtained from a high-quality analogue drum machine. It's recognisably not a real cymbal, but it does exude what you might call 'cymbal-ness', and would no doubt work well in a mix.

I think using this article as a starting point and then adjusting the envelopes and reducing the "musicality" of the sound should get you where you want to be, or at least pretty close.