Skip to content

Implement Talk Easy: on-device voice dictation app for iOS 26#1

Open
mps wants to merge 3 commits intomainfrom
claude/build-talk-easy-app-W7Fz0
Open

Implement Talk Easy: on-device voice dictation app for iOS 26#1
mps wants to merge 3 commits intomainfrom
claude/build-talk-easy-app-W7Fz0

Conversation

@mps
Copy link
Owner

@mps mps commented Feb 18, 2026

Three-layer architecture:

  • Layer 1: AVAudioEngine for microphone capture
  • Layer 2: SpeechAnalyzer (iOS 26) for real-time on-device transcription
  • Layer 3: Foundation Models (~3B LLM) for text cleanup and formatting

Core components:

  • SwiftData models: DictationMode, DictationRecord, VocabularyEntry
  • Services: AudioEngine, SpeechRecognition, TextProcessing, Clipboard
  • Multi-pass LLM pipeline with @generable guided generation
  • RecordingViewModel orchestrating the full recording flow

SwiftUI views:

  • RecordingView with animated record button and live transcript
  • ModeSelector with 7 built-in modes (General, Text, Email, Notes, etc.)
  • ModesView/ModeEditorView for custom mode management
  • HistoryView with search across past dictations
  • SettingsView with permissions, privacy info, data management

Xcode project: com.idlefusion.talkeasy, iOS 26, Swift 6, iPhone only, objectVersion 77 with file system synchronized groups.

https://claude.ai/code/session_01UoaxzhgT4YzixnKEmr2uKE

claude and others added 3 commits February 18, 2026 02:53
Three-layer architecture:
- Layer 1: AVAudioEngine for microphone capture
- Layer 2: SpeechAnalyzer (iOS 26) for real-time on-device transcription
- Layer 3: Foundation Models (~3B LLM) for text cleanup and formatting

Core components:
- SwiftData models: DictationMode, DictationRecord, VocabularyEntry
- Services: AudioEngine, SpeechRecognition, TextProcessing, Clipboard
- Multi-pass LLM pipeline with @generable guided generation
- RecordingViewModel orchestrating the full recording flow

SwiftUI views:
- RecordingView with animated record button and live transcript
- ModeSelector with 7 built-in modes (General, Text, Email, Notes, etc.)
- ModesView/ModeEditorView for custom mode management
- HistoryView with search across past dictations
- SettingsView with permissions, privacy info, data management

Xcode project: com.idlefusion.talkeasy, iOS 26, Swift 6,
iPhone only, objectVersion 77 with file system synchronized groups.

https://claude.ai/code/session_01UoaxzhgT4YzixnKEmr2uKE
…sage

- TextProcessingService: prewarm() is instance method not static; access Response.content.text not Response.text
- SpeechRecognitionService: SpeechAnalyzer requires modules: init param; SpeechTranscriber needs locale/preset; use .text (AttributedString) and .results instead of removed APIs
- Views: Use Color.accentColor instead of .accentColor for ShapeStyle conformance
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants