rocky
c07cf4db9f
Rolls back the v0.2.0 Qwen3 CoreML on-device ASR stack and replaces the
'local engine' UX with iOS 26 SpeechAnalyzer + DictationTranscriber only.
The 'Cloud polish after ASR' toggle (ProviderConfig.localModeCloudPolishEnabled)
lets users opt into a post-ASR DeepSeek round-trip from the local engine.
Defaults to off so the local engine stays genuinely local. New PolishError.missingAPIError
surfaces an inline 'fill in your key' warning when the toggle is on but the
Keychain is empty. DeepSeek preset default model bumped to deepseek-v4-flash.
Deleted:
- OSGKeyboard/ThirdParty/Qwen3Speech/ (74 files, ~16k LoC)
- OSGKeyboard/Services/ModelManager.swift (492)
- OSGKeyboard/Services/OnDeviceModelWarmup.swift (197)
- OSGKeyboard/Services/Qwen3ASRService.swift (257)
- OSGKeyboard/Services/ModelDownloadSourcePicker.swift (126)
- OSGKeyboard/Views/OnDeviceModelsView.swift (184)
- OSGKeyboard/Views/DownloadConfirmSheet.swift (96)
- OSGKeyboardShared/Models/OnDeviceModel.swift (140)
- OSGKeyboardShared/Services/OnDeviceModelStatus.swift (104)
- Qwen3ASRServiceProvider registration in OSGKeyboardApp
- Qwen3Speech package declaration in project.yml
- 5 .qwen3ASR enum / branch reference sites in HomeView, OnboardingView,
LocalEngineSettingsRows, FlowSessionManager, ASRService, EngineServiceLabel
- Two pre-existing Swift 6 strict-concurrency errors in
LiveDictationController + FlowSessionManager (the weak [weak self] in
detached-task MainActor.run blocks) that were blocking clean builds
Added:
- LocalModelsGroup: 'Built-in iOS SpeechAnalyzer' badge + 'Cloud polish
after ASR' Switch toggle
- PolishingService: honour localModeCloudPolishEnabled; new .missingAPIKey
error case with localised warning
- AppGroupStore.localModeCloudPolishEnabled (mirrored into App Group
so the keyboard extension honours the toggle during live dictation)
- SettingsView: show provider/api sections when local-mode cloud polish
is on so the user can paste a DeepSeek key
- FlowSessionManager: route through PolishingService for local + polish-on
flow; translate missingAPIKey into a polished warning
- KeyboardViewController: handle PolishingService.PolishError.missingAPIKey
in the keyboard-side live polish path
- CHANGELOG v0.2.1: documents the rollback + new toggle
- README.md / README.zh.md: engine matrix section, data flow note
Verified: xcodebuild -scheme OSGKeyboard -destination 'generic/platform=iOS Simulator'
build succeeds under SWIFT_STRICT_CONCURRENCY=complete.
160 lines
6.9 KiB
Markdown
160 lines
6.9 KiB
Markdown
# OSGKeyboard
|
||
|
||
> 按住说话，松开即得 AI 润色文字，插入任意 App 的光标处。
|
||
> 一款源码可见的 iOS 自定义键盘语音输入工具，灵感来自 [Typeless](https://typeless.com) 和 [OpenLess](https://github.com/Open-Less/openless)。
|
||
|
||
|
||
|
||
|
||
|
||
[English README](./README.md)
|
||
|
||
---
|
||
|
||
## 这是什么？
|
||
|
||
OSGKeyboard 是商业语音输入工具的免费、源码可见替代方案。它以 **iOS 自定义键盘扩展** 的形式运行，所以你可以在 **任何 App** 里使用 —— 微信、备忘录、邮件、ChatGPT、Claude、Cursor，无所不能。
|
||
|
||
1. 长按麦克风键
|
||
2. 自由说话
|
||
3. 松开 —— AI 帮你整理成干净的文字，自动插入光标处
|
||
|
||
**音频始终在设备本地转写**（iOS 26+ 使用 `SpeechAnalyzer` + `DictationTranscriber`），**只有润色后的文本** 会发到你选择的云端 LLM。**音频永不离开你的手机。**
|
||
|
||
---
|
||
|
||
## 特性
|
||
|
||
- 🎙 **按住说话**，Typeless 风格的圆形麦克风按钮
|
||
- 🧠 **端侧 ASR**（iOS 26+ `SpeechAnalyzer` + `DictationTranscriber`）
|
||
- ✍️ **AI 润色** —— 自动加结构、补标点、修正语法、可生成列表
|
||
- 🧩 **本地 + 云端润色开关** —— 本地模式默认仅在设备上识别；若 iOS 语音识别效果不理想(远场、噪声、方言)，可开启「识别后云端润色」，默认走 DeepSeek
|
||
- 🔌 **自带 API 接入** —— 兼容任何 OpenAI 兼容协议端点（OpenAI / DeepSeek / Qwen DashScope / 自建服务器 ……）
|
||
- 🔒 **隐私优先** —— 音频不离开设备；只有润色文本会发给你选择的 LLM
|
||
- 🎨 **原生 SwiftUI** —— 暗色主题、毛玻璃、约 2000 行 Swift
|
||
- 🪶 **零依赖** —— 无 SwiftPM 包、无 CocoaPods、无 Carthage
|
||
|
||
---
|
||
|
||
## 快速开始
|
||
|
||
### 环境要求
|
||
|
||
- macOS + **Xcode 16+**（推荐 Xcode 26）
|
||
- iPhone 运行 **iOS 26.0+**
|
||
- [XcodeGen](https://github.com/yonaskolb/XcodeGen)：`brew install xcodegen`
|
||
- 一个 OpenAI 兼容 API Key（[OpenAI](https://platform.openai.com/api-keys) / [DeepSeek](https://platform.deepseek.com/api_keys) / [Qwen DashScope](https://dashscope.console.aliyun.com/apiKey) 任一）
|
||
|
||
### 编译与运行
|
||
|
||
```bash
|
||
git clone https://github.com/hkgood/OSGKeyboard.git
|
||
cd OSGKeyboard
|
||
xcodegen generate # 生成 OSGKeyboard.xcodeproj
|
||
open OSGKeyboard.xcodeproj # 或命令行编译：
|
||
xcodebuild -project OSGKeyboard.xcodeproj -scheme OSGKeyboard \
|
||
-destination 'generic/platform=iOS Simulator' build
|
||
```
|
||
|
||
### 在 iOS 中启用键盘
|
||
|
||
1. 在真机/模拟器上运行 App。
|
||
2. 按 3 步引导：**启用键盘** → **允许完全访问**（麦克风 + LLM 调用必须） → **粘贴 API Key**。
|
||
3. 在任意输入框，点 🌐 切换到 **OSGKeyboard**。
|
||
4. 长按麦克风键 → 说话 → 松开。✨
|
||
|
||
> **"允许完全访问"是必须的。** 没有它，iOS 会阻止键盘使用麦克风与网络。我们**绝不记录、存储或上传你的击键** —— 见 [`PrivacyInfo.xcprivacy`](./OSGKeyboard/PrivacyInfo.xcprivacy)。
|
||
|
||
---
|
||
|
||
## 架构
|
||
|
||
```
|
||
OSGKeyboard/
|
||
├── OSGKeyboard/ # 主 iOS App（设置、Onboarding）
|
||
│ ├── Views/ # SwiftUI 屏幕
|
||
│ ├── OSGKeyboardApp.swift # @main 入口
|
||
│ ├── PrivacyInfo.xcprivacy # 隐私清单
|
||
│ └── OSGKeyboard.entitlements # App Group 声明
|
||
├── OSGKeyboardExt/ # 自定义键盘扩展
|
||
│ ├── KeyboardViewController.swift # 主体类
|
||
│ ├── Services/
|
||
│ │ ├── AudioCaptureService.swift # AVAudioEngine → 16kHz PCM
|
||
│ │ ├── ASRService.swift # iOS 26 SpeechAnalyzer ASR
|
||
│ │ └── PolishingService.swift # LLM 调用（带超时）
|
||
│ └── Views/ # 录音按钮、波形、键盘主视图
|
||
├── OSGKeyboardShared/ # 主 App + 键盘共享 framework
|
||
│ ├── Models/ # ProviderConfig、LLMRequest、LLMProvider
|
||
│ ├── Services/ # LLMClient（OpenAI 兼容）
|
||
│ └── Constants/ # App Group ID
|
||
├── OSGKeyboardTests/ # XCTest 单元测试
|
||
├── project.yml # XcodeGen 工程定义
|
||
└── .github/workflows/ci.yml # Lint + 编译 CI
|
||
```
|
||
|
||
### 数据流
|
||
|
||
```
|
||
[长按麦克风] → AudioCaptureService → AudioBufferSnapshot (16kHz mono)
|
||
↓
|
||
ASRService.transcribe()
|
||
↓
|
||
ASREvent.final(原始转写文本)
|
||
↓
|
||
PolishingService.polish()
|
||
↓
|
||
LLMClient（OpenAI 兼容协议）
|
||
↓
|
||
textDocumentProxy.insertText(润色后文本)
|
||
```
|
||
|
||
**引擎模式：**
|
||
|
||
- `cloud`（默认）：本地 `SpeechAnalyzer` 识别，文本发到云端 LLM 润色。
|
||
- `local`：本地 `SpeechAnalyzer` 识别，原始文本直接插入，不联网。
|
||
- `local` + 「识别后云端润色」开关（设置 → 本地模型）：与 `cloud` 类似，但使用本地引擎流程，识别完成后送 LLM 润色后再插入。需在 Keychain 中提前填好 DeepSeek（或任意 OpenAI 兼容提供方）API Key。
|
||
|
||
---
|
||
|
||
## 新增 LLM 提供商
|
||
|
||
打开 `OSGKeyboardShared/Models/LLMProvider.swift`，在 `presets` 数组里追加一条 `LLMProvider` 即可。默认的 `OpenAICompatibleClient` 处理任何实现了 `POST /chat/completions` 的端点。
|
||
|
||
```swift
|
||
LLMProvider(
|
||
id: "groq",
|
||
name: "Groq",
|
||
defaultBaseURL: "https://api.groq.com/openai/v1",
|
||
defaultModel: "llama-3.1-70b-versatile",
|
||
apiKeyURL: URL(string: "https://console.groq.com/keys")
|
||
)
|
||
```
|
||
|
||
仅此而已，**无需改动其他代码**。
|
||
|
||
---
|
||
|
||
## 限制
|
||
|
||
- iOS 沙盒：键盘扩展 ~60 MB 内存上限，必须开完全访问
|
||
- 密码框与部分 `WKWebView` 输入框不可用（iOS 限制）
|
||
- 仅支持 iOS 26+；更早的 iOS 版本不在支持范围内。
|
||
|
||
---
|
||
|
||
## 许可
|
||
|
||
[OSGKeyboard 源码可见许可协议](./LICENSE) —— 仅限个人学习与非商用本地使用；禁止商用、再分发及公开 fork。商业授权请联系 [rocky.hk@gmail.com](mailto:rocky.hk@gmail.com)。
|
||
|
||
---
|
||
|
||
## 致谢
|
||
|
||
- 灵感来源：[Typeless](https://typeless.com) 与桌面端开源版 [OpenLess](https://github.com/Open-Less/openless)
|
||
- 工程脚手架：[XcodeGen](https://github.com/yonaskolb/XcodeGen)
|
||
- 端侧 ASR：Apple [SpeechAnalyzer](https://developer.apple.com/documentation/speech/speechanalyzer) / [SFSpeechRecognizer](https://developer.apple.com/documentation/speech/sfspeechrecognizer)
|
||
|
||
---
|
||
|
||
**说明**：本项目发布仓库为 [`hkgood/OSGKeyboard`](https://github.com/hkgood/OSGKeyboard)。
|