Release Log
What's new in each version of Voice Decoder
v1.3.0
LatestNeural Text-to-Speech (Qwen3-TTS)
- Added Qwen3-TTS 0.6B neural TTS engine — natural-sounding speech for English and Russian
- 9 built-in voice presets (Chelsie, Ethan, and more)
- Runs fully locally on Apple Silicon via MLX — no internet required
- Generation parameters: temperature, top-p, repetition penalty, EOS bias
- Automatic text chunking by sentences for stable long-text generation
- Model pre-loading on app start for instant playback
Improvements
- Fixed race condition when switching playback between history transcriptions
- Play/Stop button state now correctly tracks the active transcription
- Word highlighting works reliably when switching between recordings
v1.2.0
AI Summarization — Built-in, No External Apps Required
- Bundled llama-server engine — no need to install Ollama or any other software
- Download and manage AI models directly from Settings
- 4 models available: GPT-oss 120B, GPT-oss 20B, Qwen 3 8B, Qwen 3 4B
- Split-GGUF support for large models (multi-file download with combined progress)
- One-click summarization of transcription results
Other Improvements
- Added Spanish localization
- Localized Sparkle update release notes (English + Russian)
v1.1.0
Text-to-Speech
- Speak transcription text and custom text aloud using macOS system voices
- Export audio to M4A with adjustable speed
- Word highlighting during playback
- TTS settings: voice selection (RU/EN), speed, pitch, volume, paragraph pause
- Personal Voice support
Improvements
- Reset TTS parameters to defaults button
- Fixed audio export error when overwriting an existing file
v1.0.0
Voice Decoder 1.0.0
- Freemium model: free tier with 10 daily transcriptions and 30s recording limit, Pro license via Gumroad for unlimited access
- License activation with Gumroad keys and promo codes
- Final retranscription for improved text quality after recording
- Segment timestamps in file transcription
- Transcription history with search and export
- Bug fixes and stability improvements
v0.4.0
Website
- Release Log page with full version history
- Download links for all previous versions
v0.3.1
Improvements
- History viewer button in settings
- Default insertion mode changed to Copy only
v0.3.0
Translation
- Translate transcription results using Apple Translation (on-device)
- Toggle between original and translated text
- Translation language configurable in Settings
Improvements
- Fixed file transcription window loading state
- Better onboarding visibility on first launch
- Fixed translation cache when changing target language
v0.2.1
App Icon
- Custom app icon matching the website design
- Icon now displays correctly in About and Sparkle dialogs
v0.2.0
Auto-Updates
- Sparkle 2.x integration for seamless updates
- "Check for Updates" in menu bar and About tab
- Automatic update checks on launch
v0.1.0
Real-Time Transcription
- Live voice-to-text powered by whisper.cpp
- Sub-second latency with Apple Neural Engine
- Voice Activity Detection (VAD)
Language Support
- Russian and English transcription
- Auto-detect mode with language switching
System Integration
- Global hotkey works in any application
- Automatic text insertion at cursor position
- Non-activating floating panel
File Transcription
- Audio: MP3, WAV, M4A, FLAC, OGG
- Video: MP4, MOV, MKV, WebM
- Batch processing with progress indication