epic: Professional Audio DSP Suite — closing the gap with Ableton/Logic/Waves/Melodyne/Ozone

[ChuxiJ]

Epic: Professional Audio DSP Suite

Vision

Transform ACE-Step-DAW's audio processing from "functional prototype" to "professional production tool" — matching or exceeding the DSP quality AND visual quality of Ableton Live, Logic Pro, and flagship plugins (Waves, Melodyne, iZotope Ozone, FabFilter). This epic tracks 12 themed feature areas that collectively close the gap.

Why This Matters

ACE-Step's unique value is AI-native music creation. But AI-generated audio still needs professional mixing, mastering, and processing to become release-ready. Without professional DSP, users must export to another DAW for final production — breaking the workflow and losing the AI-native advantage. Professional DSP keeps the entire workflow inside ACE-Step.

And beyond capability, the visual design of effects matters equally. Top plugins like FabFilter and iZotope are visual instruments that attract artists. A compressor without a transfer curve, a reverb without a decay display, a knob without texture — these signal "toy" not "tool". The UI must be as professional as the algorithms.

Design Methodology: Reference-Driven, Not Assumption-Driven

Every design decision — visual, interaction, or algorithmic — must be validated against real professional references. We do not trust our own aesthetic instincts. Instead:

1. Collect 3+ reference screenshots/recordings from professional plugins
2. Pixel-level comparison: overlay our version vs references
3. Gap list: enumerate every specific difference
4. Iterate by closing gaps, not by inventing from scratch
5. Store all references in .llm/research/ui-references/

Current State Summary

The existing effects engine has 10 insert effects (EQ3, Parametric EQ, Compressor, Reverb, Delay, Distortion, Filter, Chorus, Flanger, Phaser) plus a master chain (3-band EQ + compressor + limiter + stereo width). These are primarily Tone.js wrappers with basic parameter exposure. The AudioWorklet migration epic (#1118) provides the architectural path to run custom DSP off-thread. The UI is functional flat cards with basic SVG knobs, minimal visualization, and no visual brand identity.

Themed Issues

DSP Algorithms & Capabilities

#	Theme	Issue	Key Deliverables	Priority
1	Dynamics	#1202	Multiband compressor, gate, de-esser, transient shaper, parallel compression	P0
2	Pitch Correction	#1206	Phase vocoder, PSOLA, auto-tune, Melodyne-style manual editing, formant preservation	P0
3	EQ	#1207	Linear phase, dynamic EQ, mid/side EQ, match EQ, spectrum analyzer overlay	P0
4	Limiter & Loudness	#1208	True peak limiter, LUFS workflow, dithering, loudness-matched A/B	P0
5	Metering	#1209	Goniometer, vectorscope, correlation, LUFS meters, spectrogram	P1
6	Saturation	#1210	Tape/tube/console emulation, harmonic exciter, multiband saturation, oversampling	P1
7	Stereo Imaging	#1211	M/S processing, multiband imager, panning laws, mono bass, Haas effect	P1
8	Reverb	#1212	Dattorro plate, FDN, early reflections, convolution IR manager, ducking reverb	P1
9	Noise Reduction	#1213	Spectral de-noise, de-click, de-clip, de-hum, adaptive noise reduction	P1
10	Time-Stretch	#1214	Phase vocoder, WSOLA, material-aware modes (Beats/Complex/Texture), warp engine	P0

UI Design & Visual Quality

#	Theme	Issue	Key Deliverables	Priority
11	Design System	#1240	Professional knobs (depth/texture), unified layout grid, color language, expandable panels, A/B toggle	P0
12	Signature Visualizations	#1241	Per-effect real-time displays: compressor curve, reverb decay, delay taps, distortion transfer function, filter sweep, modulation waveforms	P0
13	Micro-Interactions	#1242	Damped knob physics, meter ballistics, bypass transitions, parameter animation, hover highlighting	P1

Dependencies

#1123 (Core DSP Library) ──→ All DSP issues (provides primitives)
#1118 (AudioWorklet Migration) ──→ All DSP issues (provides runtime)

#1202 (Dynamics) ──→ #1207 (EQ) shares multiband crossover (Linkwitz-Riley)
#1206 (Pitch) ──→ #1214 (Time-Stretch) shares phase vocoder / STFT
#1202 (Dynamics) ──→ #1208 (Limiter) shares envelope follower
#1210 (Saturation) ──→ uses oversampling from #1208 (Limiter)
#1211 (Stereo) ──→ #1207 (EQ) M/S processing shared
#963 (Spectral) ──→ #1213 (Noise) shares FFT/STFT framework

#1240 (Design System) ──→ #1241 (Visualizations) uses shared knob/layout primitives
#1240 (Design System) ──→ #1242 (Micro-Interactions) uses shared animation infrastructure
#1241 (Visualizations) ──→ Each DSP issue (visualizations depend on DSP data)

Shared Infrastructure (build once, use everywhere)

DSP Modules (from #1123)

Module	Used By
Biquad filter	EQ, Dynamics (sidechain filter), De-hum, Crossovers
Delay line	Reverb, Chorus/Flanger, Haas effect, Lookahead
FFT	Spectrum analyzer, Spectral de-noise, Phase vocoder, Match EQ
Envelope follower	Compressor, Gate, De-esser, Transient shaper, Dynamic EQ
Oversampling (polyphase FIR)	Limiter, Saturation, Waveshaper
Waveshaper	Saturation, Tape/Tube emulation, Exciter
LFO	Chorus, Flanger, Phaser, Reverb modulation, Tape wow

UI Modules (from #1240)

Module	Used By
Professional Knob component	Every effect card
EffectCardLayout grid	Every effect card
Canvas visualization framework	Every effect visualization (#1241)
Animation spring physics	Meter ballistics, knob damping (#1242)
Value tooltip component	Every knob/slider

Quality Targets

• Transparency: Processing should be inaudible when set to neutral
• Latency: All real-time effects < 10ms latency in AudioWorklet
• CPU: Each effect instance < 3% CPU at 48kHz stereo
• Visual: Every effect has a signature visualization that makes its behavior immediately readable
• Animation: 60fps for all visualizations without impacting audio
• Accessibility: All labels ≥ 10px, WCAG AA contrast, keyboard navigable

Implementation Strategy

1. Phase 1 (P0 — Foundation): Design System (feat: Effects UI Design System — Professional Knobs, Layout Grid, Color Language, Expandable Panels #1240) + Dynamics (feat: Professional Dynamics Processing Suite — Multiband Compressor, Gate, De-esser, Transient Shaper #1202) + EQ (feat: Professional EQ Suite — Linear Phase, Dynamic EQ, Mid/Side, Match EQ, Analyzer #1207) + Limiter (feat: Mastering-Grade Limiter & Loudness Management — True Peak, LUFS Workflow, Dithering #1208) + Time-Stretch (feat: Professional Time-Stretch Engine — Phase Vocoder, WSOLA, Élastique-class quality #1214)
2. Phase 2 (P0 — Visualization): Signature Visualizations (feat: Per-Effect Signature Visualizations — Real-time Curves, Meters, Waveforms for Every Effect #1241) for Phase 1 effects + Pitch Correction (feat: Pitch Correction & Vocal Tuning Engine — Melodyne/Auto-Tune class vocal processing #1206)
3. Phase 3 (P1 — Polish): Metering (feat: Professional Metering Suite — Goniometer, Vectorscope, Correlation, Spectrum Analyzer #1209) + Reverb (feat: Professional Reverb Engine — Algorithmic (Dattorro/FDN), Convolution IR Manager, Early Reflections #1212) + Stereo (feat: Stereo Imaging & Mid/Side Processing — Multiband Imager, M/S Matrix, Panning Laws #1211) + Micro-Interactions (feat: Effects Micro-Interactions & Animation — Damped Knobs, Meter Physics, Transition Polish #1242)
4. Phase 4 (P1 — Character): Saturation (feat: Saturation & Harmonic Processing — Tape, Tube, Console Emulation, Exciter #1210) + Noise Reduction (feat: Noise Reduction & Audio Restoration — Gate, De-noise, De-click, De-clip, De-hum #1213)

Each phase is independently shippable. Phase 1 alone makes ACE-Step viable for serious mixing.

Success Metrics

• Users can mix and master a complete song entirely within ACE-Step
• No need to export to another DAW for "finishing"
• Blind A/B test: ACE-Step master quality indistinguishable from Logic Pro master on same material
• Professional audio engineers can use the tools without feeling limited by the software
• New: Artists screenshot and share the UI because it looks good (visual identity = marketing)