Audio Delay

Audio Delay sits anywhere in an audio chain: source → filters → Audio Output. Connecting the Audio input splices the node into the shared AVAudioEngine graph; the cables are the engine wiring, everything runs on one clock, and the delay can be inserted mid-playback without a click.

Internally the node hosts an AVAudioUnitDelay. Time sets how long the input is held before it re-emerges — the echo spacing, in seconds. Feedback returns a percentage of the delayed signal back into the delay line: 0 gives a single echo, higher values give a decaying train of repeats, and values approaching 100 sustain almost indefinitely. Negative feedback inverts the polarity of each repeat, which subtly changes the tone of the trail as inverted copies interfere. Low Pass filters the delayed signal so each successive repeat gets darker — the analog tape-echo signature — while the dry signal stays untouched. Mix blends dry (0) against wet (1).

The Audio output re-emits the processed signal as a new engine-backed snapshot, so more filters can follow and an Audio FFT can tap the chain right after the delay — the analyzer then sees the echoes too. The Level output is a peak-decay RMS of the wet+dry result, which keeps moving during echo tails even after the source stops. Enable = false engages the audio unit's clean bypass: dry audio keeps flowing, unprocessed.

Outputs: Audio (the processed signal, chainable) and Level (peak-decay RMS of the processed signal — it tracks the echo tail, not just the source).

For tempo-locked echoes, compute Time from BPM with Math nodes: a quarter note is 60 / BPM seconds, a dotted eighth is 45 / BPM.

• Dub echo throw: Audio Player → Audio Delay (Time 0.4, Feedback 65, Low Pass 3000, Mix 0.5) → Audio Output — dark repeats that melt into the mix; ride Mix on a fader to "throw" phrases into the echo.
• Slap-back voice: Audio Input (mic) → Audio Delay (Time 0.08, Feedback 0, Mix 0.3) → Audio Output — the fifties vocal thickener, working on live input via the engine's source bridge.
• Echo-reactive visuals: Audio Player → Audio Delay → Audio Output, with an Audio FFT tapped on the delay's Audio output — the spectrum pulses with each repeat, so visuals keep dancing after the source goes silent.
• Tempo-synced rhythm: BPM value → Math (60 / BPM) → Audio Delay Time, in a chain Audio Player → Audio Delay (Feedback 55) → Audio Output — echoes land exactly on the beat.
• Self-oscillating drone machine: Audio Player → Audio Delay (Feedback 95, Low Pass 2000, Mix 0.8) → Audio Output — feed it a short sound, then sweep Time slowly for pitch-warping ambient beds.
• Dark tail on highs only: fan the source into Audio Filter (High Pass 2000) → Audio Delay → Audio Output alongside the dry branch — echoes only on the sparkle, bass stays tight.

• Feedback safety: at 90–100% the delay approaches self-oscillation and levels can build with every repeat. Approach the top of the range gradually, keep Mix moderate while experimenting, and consider an Audio Compressor after the delay as a safety net. Enable = false instantly bypasses if a trail runs away.
• Changing Time while audio runs re-times the delay line and can smear or click the current tail — sweep it deliberately as an effect, or change it between phrases for clean tempo switches.
• Lower Low Pass (2000–5000 Hz) makes big feedback settings musical: each repeat loses highs, so trails sink into the background instead of stacking into harshness.
• Delay before reverb sounds like echoes in a room; reverb before delay sounds like a room being echoed. Both are valid — pick per patch.
• Negative Feedback is subtle on most material but audibly changes short slap-backs and metallic timbres — worth an A/B when a repeat sounds too "in your face".
• Mix at 1.0 with Feedback near 0 outputs only the delayed copy — a clean way to build a deliberate pre-delay or latency offset into a parallel branch.