56d0da0a519be10cb34d980d6fad1d5324974238
3 Commits
| Author | SHA1 | Message | Date | |
|---|---|---|---|---|
 Rocky
|
a227309059 |
fix: feed DictationTranscriber Int16 PCM, not Float32
The keyboard preview crashed on first record with a
`__abort_with_payload` deep inside Speech's
`DictationTranscriber`. The disassembly surfaced three
preconditions checked before a `brk #0x1`:
+620 "Audio sample data must be 16-bit signed integers"
+848 "Multi-channel audio is not supported"
+1072 "Client info not fully initialized"
We hit the first one. `DictationTranscriber` (iOS 26's new
`SpeechAnalyzer`-backed engine) is strict about its input
format: only Int16 PCM, not the Float32 PCM that the iOS 18
`SFSpeechRecognizer` path accepted. Our audio-tap and
`AudioBufferSnapshot.samples: [Float]` are Float32 all the
way down — that was the SFSpeech shape, and the previous
`AppleSpeechASR` adapted internally. With iOS 26 as the
deployment target, the only ASR backend is
`SpeechAnalyzerASR`, and the conversion needed to happen at
the `AnalyzerInput` boundary.
Fix:
- `transcribe` builds the `AVAudioFormat` as
`.pcmFormatInt16, 16 kHz, 1 ch, interleaved: true` (the
canonical layout for Int16 Speech input).
- `makeInputStream` runs the per-sample conversion
`Int16(round(clamp(s * 32767, -32768, 32767)))` into the
`AVAudioPCMBuffer`'s `int16ChannelData[0]`. The explicit
clip is required (a `s == 1.5` from a gain-overflow at the
audio-engine boundary would otherwise wrap to a negative
Int16 after the implicit truncation). `round()` (not
truncate) preserves DC balance — `0.5` quantises to
`+16384`, not `+16383`, matching what audio DAWs expect.
- The conversion helper is exposed as
`ASRServiceFactory.convertFloat32ToInt16` so unit tests
can lock the math without instantiating the full pipeline.
Why not change `AudioBufferSnapshot` to `[Int16]` instead
(see earlier first-principles discussion): the snapshot is a
transport format that both `AudioCaptureService` (in the
ext) and `PreviewASRController` (in the main app) produce.
Float32 is the natural shape coming out of `AVAudioEngine`,
and pushing the conversion to the ASR service keeps the
transport contract platform-agnostic — a future second
backend with different format needs can have its own
adaptation without dragging everyone else.
Tests:
- `testFloat32ToInt16EdgeCases` — 0, ±1, ±0.5, ±1.5
(gain-overflow case).
- `testFloat32ToInt16RoundTrip` — quantisation step is
1/32767 (so the asymmetric Int16 range is honoured: -32768
has no exact Float source).
- `testFloat32ToInt16Empty` — `sourceCount == 0` with nil
pointers is a no-op (function guards on count before
dereferencing).
- All 25 tests pass (22 existing + 3 new).
- BUILD SUCCEEDED.
🤖 Generated with Claude Code
|
||
|
Rocky
|
81581f0e5f |
chore: drop iOS 18–25 / non-iPhone support, require iOS 26
Two related cleanups the user asked for in one shot:
1. iPhone-only is now enforced at every target — Mac Catalyst and
visionOS were never configured in `project.yml`, but the
`OSGKeyboardShared` framework and the two test bundles were
still defaulting to `TARGETED_DEVICE_FAMILY = "1,2"` (iPhone +
iPad). All four targets now explicitly set `"1"`. SDK is
`iphoneos` for everyone, no `xros` / `macosx`.
2. Deployment target bumped from iOS 18.0 to iOS 26.0 across the
board (`project.yml` + the ext's per-target setting). With
iOS 26 as the floor, the iOS 18–25 SFSpeechRecognizer path
became dead code and several `#available` checks became
always-true. Removed:
- `AppleSpeechASR` (the entire SFSpeechRecognizer-based ASR
backend) and the `#available(iOS 26.0, *)` factory branch.
`ASRServiceFactory.make()` now returns `SpeechAnalyzerASR()`
directly. SpeechAnalyzer is always fully on-device, which
also made the `requiresOnDevice` flag meaningless.
- `requiresOnDevice` from the `ASRService.transcribe` protocol
signature, from `ProviderConfig`, `AppGroupStore`,
`KeyboardState`, `AppGroupPersistor`, and the ext's
`KeyboardViewController` (`state.requiresOnDevice`,
`state.setRequiresOnDevice`, `persistRequiresOnDevice`).
- `#available(iOS 17.0, *)` branch in
`PreviewASRController.requestMicrophonePermission` and the
ext's `PermissionManager.requestMicPermission` — both now
just call the iOS 17+ `AVAudioApplication` API directly.
- The `else` (iOS 18–25) branch in `SettingsView.asrEngineRow`
— the on-device-only toggle is gone, the row is a static
"SpeechAnalyzer active" badge. Same for the `else` branch
in `EnginePickerSection.localSubtitle`.
- `makeMicAuthHandler` (the iOS < 17 mic permission callback
wrapper) from `PreviewASRController`.
No `#available` / `@available` checks remain in the codebase
except for the SpeechAnalyzer class itself (now unnecessary
too, but kept for clarity — `AVAudioApplication` and
`SpeechAnalyzer` are both iOS 17+ / iOS 26+ respectively,
and the deployment target of 26 makes the explicit
`@available` redundant; I left the SpeechAnalyzer class
un-`@available` and removed the `@available(iOS 26.0, *)`
decoration since it's no longer needed).
The Keyboard ext's existing ASRService usage
(`asr.transcribe(stream:locale:)`) is unchanged at the
call-site level — just the third argument is gone.
3. Updated `info.plist` UISupportedInterfaceOrientations is
already `[UIInterfaceOrientationPortrait]` only, which is
correct for an iPhone-only app; no change needed.
Build: BUILD SUCCEEDED.
Tests: 22/22 pass.
Verified: `TARGETED_DEVICE_FAMILY = 1` on all four targets,
`SDKROOT = iphoneos` on all four.
🤖 Generated with Claude Code
|
||
|
Rocky
|
aeb28f4b36 |
fix: green accent + real iOS ASR in preview + bilingual audit
Three user-flagged fixes, scoped tightly to the files each affects.
The uncommitted Engine-mode wiring / locale picker / etc. from a
prior agent pass is intentionally not included in this commit.
1. Revert accent to brand green (#3AA05A).
Last commit flipped `AccentColor` + `Palette.light.accent` to
Apple system blue (#007AFF) on the assumption that "green CTA
on a near-white surface looks wrong." Review pushed back:
the brand *is* the green, and the system tint should match it
so that NavStack Done buttons, Toggles, and our custom
`primaryButton()` modifier all read as the same colour. Restored
`#3AA05A` in both `AccentColor.colorset/Contents.json` and
`Palette.light.accent` (plus the muted / glow variants).
2. Real iOS ASR in `KeyboardPreviewSheet`.
The previous fix only swapped the static placeholder for a
`TextField` and routed a hardcoded stub through it — review
asked, fairly, "are you actually calling `SFSpeechRecognizer`?"
Answer: no. This change makes the preview *run real ASR*:
- `ASRService` (+ the iOS 26 `SpeechAnalyzer` path) moves from
`OSGKeyboardExt/Services/` to `OSGKeyboardShared/Services/`,
so the host app can import the same `ASRServiceFactory.make()`
the keyboard extension uses.
- `OSGKeyboardShared` gains `Speech.framework` and
`AVFoundation.framework` as SDK dependencies in `project.yml`.
- New `OSGKeyboard/Views/PreviewASRController.swift` (the
extension's `AudioCaptureService` is app-extension-only, so
the preview owns its own `AVAudioEngine` + `AVAudioSession`
and downsamples to 16 kHz mono Float32 via `AVAudioConverter`).
- `KeyboardPreviewSheet.cyclePhase()` now calls
`asr.start(locale:)` / `asr.stop()` instead of toggling a
`StubPhase`. The disc's level meter is driven by RMS from the
actual audio tap; the transcript line under the chips shows
the live `SFSpeechRecognizer` partial; the top textbox
receives the `.final` transcript via `onChange(of: lastFinal)`.
3. Record disc in BOTH stubs now uses the green accent for idle.
Was `Color(white: 0.22)` dark-gray, which read as "inert
surface" rather than "tap me". The keyboard extension and the
in-app preview now share the same brand-green disc gradient so
the keyboard's primary CTA is the same colour in both modes.
4. Bilingual audit across every user-facing string.
Every Text() in the main-app and extension views is now
`中文 · English` or carries an English secondary line. Covered:
- OnboardingView (Back, Next, Done, Continue, step instructions,
PrivacyFootnote rows)
- HomeView (status header, hero label, accessibility)
- APISettingsCard (Base URL, API Key, Model, "Get an API key",
"Connection", test-connection states + error messages)
- KeyboardPreviewSheet (title, subtitle, TextField placeholder,
clear button accessibility)
- KeyboardPreviewStub (mode/locale chips, REC badge, space bar)
- KeyboardRootView (gear accessibility, space bar, requesting
state, denied messages, local-engine chip)
- RecordButton (accessibility label)
- SettingsView (Done, Reset dialog, language section subtitle,
engine section, local-engine subtitle, on-device label)
- AppGroupErrorView (title, body, three remediation steps)
- LLMProvider preset names and blurbs
Where the prior pattern was "Chinese headline + English footnote"
(e.g. OnboardingView's 启用 OSGKeyboard / Enable OSGKeyboard),
that pattern was preserved — bilingual coverage means every
screen reads as both, not that every line is rigidly `中 · EN`.
Build: BUILD SUCCEEDED on iPhone 17 Pro / iOS 26 simulator.
Tests: 21/21 pass (no test changes).
Visual: light-mode home shot at /tmp/osgk_light_green.png shows
the brand-green CTA restored across Next button, mic icon, and
page dot.
🤖 Generated with Claude Code
|