soundcraft/docs/target-app-parity.md

# Target-app parity: SoundCraft vs Avid Pro Tools

> **Last reviewed:** 2026-10-10 · **Last updated:** 2026-10-10 · **Change:** minor (hours to ~95 % for each audience; full number confirmed as an additive weighted sum) · **Target:** Avid Pro Tools Ultimate 2026.4.1 (bundle 26.4.1.179, macOS)

The authoritative assessment of how close SoundCraft is to Pro Tools. [`ROADMAP.md`](../ROADMAP.md)
summarizes it, [`gaps.md`](gaps.md) itemizes every shortfall, and the per-dimension docs go deeper:
[`ui-parity.md`](ui-parity.md), [`file-format-parity.md`](file-format-parity.md),
[`hardware-parity.md`](hardware-parity.md), [`plugin-parity.md`](plugin-parity.md),
[`localization-parity.md`](localization-parity.md). The generated menu checklist is
[`parity-checklist.md`](parity-checklist.md).

## Headline

| Number | Value | Kind |
|---|---|---|
| Menu breadth (engine + UI layer) | **482 / 512 menu leaves (94.1 %)** | **measured**: the 512-leaf catalog `crates/engine/catalog/menus.txt` against the command registry; engine alone 397 / 512 (77.5 %) from `cargo xtask parity` ([`parity-checklist.md`](parity-checklist.md)); the 85 UI-layer matches re-counted on 2026-10-10 from `crates/ui-egui/src/menus.rs` (`UI_COMMANDS`, `AUDIOSUITE`, `MENU_WINDOWS`, Mix Window Views) by script |
| Feature breadth (menus plus everything not in a menu: plugins, hardware, formats, immersive, cloud) | **~70 %** | estimated (area table below, presence only) |
| **Ready for real work** (whole of Pro Tools) | **~40 %** | estimated: written dimension weights × values = 40.6 % ([By dimension](#by-dimension)) |
| **Mainstream practitioner** (a working tracking/mixing engineer) | **~36 %** | estimated: written area weights and discounts ([below](#mainstream-practitioner-and-essentials-user)) |
| **Essentials user** (making a song or podcast with the core features) | **~57 %** | estimated: written feature weights and discounts ([below](#mainstream-practitioner-and-essentials-user)) |
| Stage | **pre-alpha** (core-workflow gate fails) | see [Stage](#stage) |
| Remaining effort to alpha | **~110–180 Opus 5.5 agent-hours** | estimated: the failing gate rows |
| Remaining effort to beta | **~350–570 Opus 5.5 agent-hours** | estimated, calibrated (below) |
| Remaining effort to full parity | **~950–1,450 Opus 5.5 agent-hours** | estimated, calibrated (below) |

### Readiness by audience

| Audience | Ready % | Opus 5.5 agent wall-clock hours to ~95 % | Work that dominates |
|---|---:|---:|---|
| Full target (ready for real work) | ~40 % | ~950–1,450 h (~60–70 % parallelizable) | Hardware (interfaces, MIDI, surfaces, sync), the built-in plugin catalogue, AAF/OMF and immersive delivery, localization, fidelity against Pro Tools |
| Mainstream practitioner | ~36 % | ~380–600 h (~60 % parallelizable) | Interface recording and MIDI devices, AAF/OMF exchange, stock-plugin depth, the core-path bug backlog, interaction feel |
| Essentials user | ~57 % | ~110–200 h (~70 % parallelizable) | Launch and platform fixes, an onboarding and user guide, stock-effect panels and presets, MP3/AAC export, editing and transport bugs |

Each audience's hours are for its own areas only, so essentials ⊂ mainstream ⊂ full. Calibration
is the same as for the full estimate ([Effort and calibration](#effort-and-calibration)): bug fixes
at 0.5–1.5 h each, features at about 200–300 tested lines per agent-hour, from the public-week PRs.
- **Mainstream sum:** recording 40–60, MIDI 50–80, editing 40–60, mixing 35–50, automation 20–30,
  stock plugins 80–120, AAF/OMF 50–90, compressed export 10–20, bug backlog 25–40, interaction
  feel 30–50.
- **Essentials sum:** default-device recording 10–15, transport 5–10, editing polish 15–25, effect
  panels and presets 20–30, MP3/AAC export 10–20, built-in instrument 10–20, launch and platform
  15–25, onboarding and user guide 15–25, save fixes 5–10.

The previous figure in ROADMAP.md (2026-10-07: "~65 % overall feature parity", "110–150 h to 100 %")
measured feature *presence*. This pass measures readiness, and finds new evidence that pulls the
numbers down: 37 open GitHub issues from the first public week (about 25 of them correctness bugs
in features counted as done), no MIDI hardware input or output at all, no audio device or buffer
size selection, no localization, and a macOS release that cannot load signed third-party VST3s
(#153). The estimates did not drift up; the scope of what is measured widened.

## How this was measured

- **The target.** `/Applications/Pro Tools.app` on the owner's Mac: Info.plist reports version
  26.4.1.179 (Pro Tools Ultimate 2026.4.1). Only bundle metadata was read (version,
  `CFBundleDocumentTypes` for the file types it opens, the `.lproj` folder names for its UI
  languages: en, de, es, fr, ja, ko, zh-Hans, zh-Hant). Nothing inside the bundle's resources,
  plug-ins or code was read or copied (clean-room rule, AGENTS.md).
- **The menu tree.** `plan/protools/menus-clean.txt` (observed black-box, local only) became the
  committed 512-leaf catalog; `cargo xtask parity` and the UI layer's `engine.parity` score it.
- **Avid's own documentation.** *What's New in Pro Tools 2026.4* (Track Pin, MPEG-H renderer,
  Dolby headphone personalization, speech-to-text upgrades, batch file rename) and earlier
  release notes for features outside menus (Dolby Atmos renderer integration, ADM BWF, Sketch,
  cloud collaboration, Splice, ARA, HEAT, control surfaces, I/O setup).
- **The code on origin/main (3bc2d92, 2026-10-10).** Read and counted with grep: ~70,900 lines of
  Rust; 524 `#[test]` functions plus proptest suites; ~350 macro-registered engine command ids;
  26 built-in processors and 2 instruments in `crates/dsp/src/plugins`; audio I/O in
  `crates/audio-io`; device I/O in `crates/playback` (cpal default output and default input only);
  no MIDI device crate (no `midir`/CoreMIDI anywhere); no string catalog.
- **Users.** The 37 open and 21 closed GitHub issues, and the 42 merged and 29 open pull
  requests up to 2026-10-10, are the only field evidence. Nothing has been benchmarked against
  Pro Tools.
- **Not done:** running Pro Tools side by side on the same material (null tests of mixes, plugin
  responses, fades, elastic audio). That is gap G-01 in [`gaps.md`](gaps.md).

## By dimension

Weights are how much each dimension decides whether a Pro Tools user (music production,
post-production, tracking) can switch. "Ready" is depth, correctness and fidelity, not presence.

| Dimension | Weight | Ready | Kind | Remaining (h) | Doc | Evidence |
|---|---:|---:|---|---:|---|---|
| Features | 45 % | **~45 %** | estimated | 520–810 | [area table](#feature-areas) | Menu breadth 94 % measured, but depth is thin: see the area table |
| UI/UX fidelity | 10 % | **~55 %** | estimated | 60–100 | [ui-parity.md](ui-parity.md) | Edit/Mix windows, every menu, floating windows; open UX bugs #3 #5 #6 #7 #140 #144; shortcuts only recently right on Windows/Linux (#15, #41); generic plugin UIs |
| File formats | 10 % | **~40 %** | estimated | 70–120 | [file-format-parity.md](file-format-parity.md) | WAV/BWF/RF64/AIFF/FLAC read+write, compressed formats read; no `.ptx`, AAF, OMF, ADM BWF, MP3/AAC export, video export |
| Hardware | 10 % | **~12 %** | estimated | 110–180 | [hardware-parity.md](hardware-parity.md) | Default output and default input device only, fixed 512-sample block, no ASIO, no MIDI devices, no control surfaces, no sync |
| Stability | 10 % | **~45 %** | estimated | 50–80 | [gaps.md](gaps.md) | Never-crash rules and proptests (random edits, mutated sessions, hostile params); but ~25 open correctness bugs, a launch failure (#142), an X11 llvmpipe failure (#107) |
| Performance | 5 % | **~35 %** | estimated (unmeasured) | 40–70 | [hardware-parity.md](hardware-parity.md) | Parallel strip rendering and PDC exist; whole files decoded into RAM, recording into RAM (#84 open), no track-count or latency benchmarks |
| Ecosystem / plugins | 4 % | **~30 %** | estimated | in Features + Hardware | [plugin-parity.md](plugin-parity.md) | CLAP/VST3/AU hosting; no AAX (Pro Tools' only format), no ARA, no sidechain, no control-surface ecosystem, no content library |
| Platforms | 3 % | **~65 %** | estimated | 20–40 | [ROADMAP.md](../ROADMAP.md) | Builds for macOS, Windows x64/x86/arm64, Linux, FreeBSD, Web (Pro Tools: macOS, Windows), but only macOS is exercised at runtime; Web lacks file pickers (#63, #64) |
| Localization | 2 % | **~12 %** | measured (catalog count) | 70–110 | [localization-parity.md](localization-parity.md) | English only, no string catalog; Pro Tools ships 8 UI languages; FR/ES PR #143 open |
| AI features | 1 % | **~5 %** | estimated | 30–50 | [gaps.md](gaps.md) | Pro Tools has speech-to-text transcription (upgraded in 2026.4); we have only an audio-to-MIDI pitch tracker |
| **Weighted** | 100 % | **~40 %** | | **~950–1,450** (overlaps removed) | | |

Agent control is not weighted (Pro Tools has only its scripting SDK); SoundCraft is ahead there:
every command is reachable from the CLI, the JSON control channel and MCP.

## Mainstream practitioner and essentials user

Both use the method in craftrules `standards/progress-docs.md`. Each multiplies a weighted depth
average by written-down discounts. Depths come from the [Feature areas](#feature-areas) table and
the evidence there; issue numbers are `storytold/soundcraft` GitHub issues.

### Mainstream practitioner: ~36 %

The typical Pro Tools professional is a music or post engineer who tracks through an audio
interface, overdubs MIDI, edits, mixes with the stock plugins and bounces every week. This number
leaves out third-party plugin hosting, Avid cloud services and AI, control surfaces and sync,
immersive delivery, video, notation, and languages other than English.

| Area (weekly use) | Weight | Depth |
|---|---:|---:|
| Recording and monitoring through an interface | 20 % | 40 % |
| Editing (trim, fades, comping, Slip/Grid, nudge) | 22 % | 65 % |
| Mixing and routing (faders, sends, busses, inserts, PDC) | 18 % | 60 % |
| Automation | 10 % | 55 % |
| Built-in plugins and AudioSuite | 12 % | 35 % |
| MIDI recording, editing, instruments | 10 % | 30 % |
| Bounce, save and reopen | 8 % | 60 % |
| **Weighted depth** | 100 % | **50.6 %** |

| Discount | Factor | Evidence |
|---|---:|---|
| Interaction fidelity | ×0.90 | Shortcut collisions and dead shortcuts (#140 Heal, #144 Mix/Edit vs zoom, #7 Export); Multitool handles don't adjust audio (#5); generic plugin parameter panels (#138, #139); clip deletion inconsistencies (#3, #4) |
| Stability on real machines | ×0.92 | 26 open core-path bugs, e.g. bounce and latency (#59, #103), playback speed (#101), session saves (#86, #91), plus a launch failure (#142). Never-crash rules and property tests hold up: no crash reports on macOS since #16 |
| File exchange with Pro Tools users | ×0.85 | Engineers trade sessions, not stems. No `.ptx`, AAF or OMF (#40); WAV/BWF stems and MIDI files do work, so the factor isn't lower |
| **Result** | 50.6 × 0.90 × 0.92 × 0.85 = **35.6 % ≈ 36 %** | |

This comes out *below* the full number (~40 %), which the standard says is unusual. The
mainstream engineer's weekly core is interface recording and MIDI overdubs, exactly the
alpha-gate failures. The full number gets partial credit from areas this user doesn't weight:
editing breadth, five platforms built, agent control.

**User evidence** (GitHub, 2026-10-10; all 69 issues are from 28 outside reporters, none from
maintainers):
- **Praise:** 0 threads, and no "switched from Pro Tools" reports.
- **Open issues:** of 48 open, 26 are core-path bugs (editing, bounce, playback, sessions,
  shortcuts, plugin loading, recording monitoring), 7 are niche-path bugs (web build, CLI/MCP,
  Linux fonts and llvmpipe), and 15 are requests.
- **Requests:** several ask for other DAWs' feel: Ableton shortcuts and layout (#71, #72), an
  FL/Ableton port (#38), Audition-style file editing (#81, #82). AAX (#109), VST2/LV2 (#30),
  Ukrainian (#67).
- **Closed:** 21.
- **Contributions:** 12 outside contributors have merged PRs in the first week. That's
  engagement, not yet use on real projects.

### Essentials user: ~57 %

A user who needs only the essentials: makes a song, demo or podcast, records through the computer's default
microphone or interface, uses default settings, and exports a file to share.

| Core feature | Weight | Depth | Notes |
|---|---:|---:|---|
| New session, open a session, demo | 10 % | 80 % | |
| Import audio (drag and drop, WAV/MP3/AAC/FLAC) | 10 % | 80 % | Reads every common format |
| Record from the default device | 12 % | 65 % | Works on the default input; forced monitoring can feed back (#42) |
| Play, stop, loop, scroll, zoom | 10 % | 75 % | Half-speed after Stop (#101) |
| Basic editing: select, cut/copy/paste, split, trim, move, fades | 18 % | 70 % | #3, #5 |
| Undo/redo | 5 % | 90 % | Every command undoable |
| Volume, pan, mute, solo | 10 % | 85 % | |
| Add a stock effect (EQ, compressor, reverb) and pick a preset | 10 % | 60 % | Generic panels, few presets |
| Draw MIDI notes for a built-in synth | 5 % | 50 % | Two built-in instruments |
| Save and reopen | 5 % | 80 % | Web build can drop unsaved edits (#136) |
| Export / bounce a file to share | 5 % | 60 % | WAV/AIFF/FLAC only, no MP3/AAC |
| **Weighted depth** | 100 % | **72.4 %** | |

| Discount | Factor | Evidence |
|---|---:|---|
| Launch and stability | ×0.90 | Doesn't launch for one user (#142), X11 llvmpipe failure (#107), Linux symbols as boxes (#45), Windows shortcuts (#41); macOS is solid |
| Discoverability and UI clarity | ×0.90 | Pro Tools' dense layout with no onboarding or user guide; shortcut collisions; plugin windows are bare sliders |
| Opening files people send them | ×0.97 | These users get WAV/MP3, which work; only Pro Tools sessions fail |
| **Result** | 72.4 × 0.90 × 0.90 × 0.97 = **56.9 % ≈ 57 %** | |

## Feature areas

Weights by what Pro Tools users actually spend time on. **Breadth** = the feature exists;
**Ready** = it works deeply and correctly enough for paid work.

| Area | Weight | Breadth | Ready | Remaining (h) | What's there / what's missing |
|---|---:|---:|---:|---:|---|
| Recording and monitoring | 12 % | 75 % | **40 %** | 40–60 | Punch in/out, QuickPunch-style punch on the fly, pre/post-roll, loop record into playlists, input monitoring through the strip, autosave/recovery. Missing: input/output device choice, buffer size, duplex low-latency path, record straight to disk (#84 open), per-channel hardware input routing, TrackPunch/destructive punch beyond flags, monitoring opt-out (#42) |
| Editing | 15 % | 92 % | **65 %** | 40–60 | Slip/Shuffle/Spot/Grid, every Edit menu command, fades/crossfades, playlists and comping, clip groups, nudge, Tab to transient, separate/heal/consolidate/strip silence. Open bugs: #3, #5 (Multitool handles), #140 (Heal shortcut), #86, #91, #92; no batch fade presets, no field-recorder workflows |
| Mixing and routing | 12 % | 85 % | **60 %** | 35–50 | Faders, pan, sends, busses, aux, masters, VCAs, folders, 10 inserts, PDC, solo modes, I/O Setup. Missing: sidechain inputs, hardware-channel I/O mapping, bus/AFL/PFL paths verified, HEAT; bugs #59, #103 |
| Automation | 7 % | 80 % | **55 %** | 20–30 | Lanes for volume/pan/mute/sends/plugin params, trim, Write/Touch/Latch, glide/thin/coalesce. Missing: surround-pan and object automation, Touch/Latch prime and preview modes, automation on control surfaces; bug #104 |
| Built-in plugins and AudioSuite | 8 % | 40 % | **35 %** | 80–120 | 26 processors + 2 instruments covering every AudioSuite category; Pro Tools bundles roughly three times as many plugins plus virtual instruments. No fidelity comparison, generic parameter UIs (#139). See [plugin-parity.md](plugin-parity.md) |
| Third-party plugin hosting | 8 % | 70 % | **45 %** | 40–60 | CLAP, VST3, AU: audio, params, latency, notes, state, presets, off-thread lifecycle, CLAP floating GUIs, VST3 editors on macOS. Missing: AU editors, VST3 editors on Windows/Linux, sidechain buses, ARA, out-of-process sandboxing, macOS library-validation entitlement (#153, #149). AAX cannot be hosted (Avid SDK licence, PACE signing) |
| MIDI and instruments | 8 % | 70 % | **30 %** | 50–80 | Piano roll, event list, step input, quantize/transpose, real-time properties, SMF import/export. Missing: **any MIDI device input/output** (no keyboard recording, no external synths, no MIDI clock/MTC out), MIDI learn, instrument depth (Sampler PR #108 open), bug #98 |
| Score and notation | 2 % | 60 % | **35 %** | 20–30 | Grand-staff Score Editor, MusicXML export, printable SVG. Missing: notation editing depth (tuplets, articulations, lyrics, parts layout) |
| Elastic Audio, TCE, Beat Detective, pitch | 5 % | 70 % | **35 %** | 40–60 | Pitch-preserving warp on Elastic tracks, TCE, conform to tempo, Beat Detective and Identify Beat, audio-to-MIDI. Missing: per-algorithm choice (polyphonic/rhythmic/monophonic/varispeed/X-Form), warp markers UI depth, Elastic Pitch, quality vs Pro Tools unmeasured |
| Surround and immersive | 5 % | 55 % | **35 %** | 50–80 | Formats stereo → 9.1.6 and Ambisonics, surround panner, fold-down, multichannel bounce, object/bed routing. Missing: Dolby Atmos renderer integration and ADM BWF, MPEG-H (new in 2026.4), binaural/headphone monitoring, surround automation; bug #89 |
| Video and post | 5 % | 60 % | **40 %** | 40–60 | Video track, H.264/ProRes/MJPEG decoders, sync offset, relinking, timecode rulers. Missing: HEVC/DNx/AV1, video bounce/export, timecode sync to external machines, speech-to-text transcript, field-recorder/iXML conform |
| Session, file management, interchange | 10 % | 55 % | **35 %** | 60–100 | `.scraft` + Audio Files, Save Copy In, Import Session Data, clip groups, session text export, audio-health report. Missing: `.ptx` read/write, AAF/OMF (#40), session templates, track presets, workspace/Soundbase browser, disk streaming of long sources, batch rename; bugs #86, #91, #136 |
| Collaboration, cloud, Sketch, Splice | 2 % | 0 % | **0 %** | not estimated | Avid services; out of scope unless the owner decides otherwise |
| Agent control (CLI, control channel, MCP) | 1 % | 100 % | **100 %** | 3 | Ahead of Pro Tools' scripting SDK; one open bug (#145) |
| **Weighted** | 100 % | **~70 %** | **~45 %** | **~520–810** | |

## Stage

**Pre-alpha.** Readiness (~40 %) is at the alpha bar, but the core-workflow gate
([`roadmap.md`](roadmap.md#alpha-gate)) fails three of six Pro Tools workflows on macOS, the main
platform:

- Tracking from a chosen interface at a chosen buffer size: **no**. Only the default devices are
  used, the mix runs in fixed 512-sample blocks, input and output are separate streams, and
  takes are held in RAM.
- MIDI recording: **no**. There is no MIDI device layer.
- Session interchange with Pro Tools or picture editors: **no**. No AAF/OMF/`.ptx`.

Editing, mixing with plugins, bouncing, and saving and reopening pass (with known bugs).

**Distance to alpha: ~5 points of readiness and ~110–180 agent-hours** (G-01, G-02, G-03, part
of G-06, and the #153 entitlement). **Distance to beta: ~35 points and ~350–570 agent-hours**
(beta also needs ~75 % ready). The beta plan is in [`roadmap.md`](roadmap.md).

## Effort and calibration

Hours are **Opus 5.5 agent wall-clock hours**: one agent session working sequentially, including
tests and verification.

**Calibrated against this repo's history** (`git log`, `gh pr list`):

- The 10-05 bulk commits (`first commit` through `Video track, Audio Unit hosting…`, ~70k lines)
  carry batch-commit times, not authoring times, so they don't calibrate.
- The public week (2026-10-07 → 10-10, 42 merged PRs) does. Focused bug fixes with a regression
  test, +20 to +120 lines (#116–#122, #25, #28, #37, #61), took **~0.5–1.5 h** each. Small
  features of +200 to +1,000 lines (#32 light theme +516, #11 log file +981, #8 credits +666, #23
  plugin menus +259, #76 clip-gain fader +198) took **~2–5 h**: about **200–300 tested lines per
  agent-hour**.
- Areas were sized by the code they need at that rate (for example, MIDI device I/O plus MIDI
  recording ≈ 3–5k lines ≈ 12–25 h), plus a 1.3–1.5× tail factor for verification against Pro
  Tools behaviour, as FilmCraft observed for its long tail.

**Parallelism.** About 60–70 % of the work parallelizes across 4–5 agents by crate (dsp plugins,
hardware/playback, file interchange, UI, localization). The audio-thread and hardware work
serializes more, because it needs hands-on testing.

**Needs a human:**
- Owner decisions: AAX (requires Avid's SDK licence and PACE signing; probably never), `.ptx`
  reverse engineering (clean-room implications), Avid cloud features (out of scope by default).
- Hardware we don't have in CI: multichannel interfaces, ASIO drivers, control surfaces (HUI/MCU,
  EUCON is Avid-proprietary), MIDI keyboards, sync generators.
- Native-speaker review for every language beyond English.
- Ears: a mix engineer's A/B of plugins, fades and elastic audio.

## Revision history

| Date | Change | Summary |
|---|---|---|
| 2026-10-10 | minor | Readiness-by-audience table: hours to ~95 % for full (950–1,450 h), mainstream (380–600 h) and essentials (110–200 h). The full number is already the additive weighted sum over the written dimension table (40.6 %); method unchanged |
| 2026-10-10 | minor | Added the mainstream-practitioner (~36 %) and essentials-user (~57 %) numbers with written weights, discounts and user evidence. Checked the full ~40 %: it is the written dimension weights × values (40.6 %), so it doesn't move |
| 2026-10-10 | minor | Stage alpha → pre-alpha under the core-workflow gate; alpha distance ~110–180 h |
| 2026-10-10 | major | Full re-measure against Pro Tools Ultimate 2026.4.1: two numbers (breadth ~70 %, ready ~40 %), dimension and area tables with hours, stage alpha; hours re-calibrated from PR history (110–150 h → 950–1,450 h to full parity: the old figure counted presence only) |
| 2026-10-07 | major | Area table and "~65 % overall, 110–150 h" estimate (then in ROADMAP.md) |