Voice Typer v1.2.0

What's new in v1.2.0

Traditional Chinese UI 繁體中文介面

• The whole main window can now switch between English and Traditional Chinese (Taiwan).
• Toggle button at the top-right of the header — shows 中文 when in EN, EN when in 中. One click flips every visible label.
• Translated: section titles, mode names, all buttons, hotkey descriptions, "Latest Transcript" / "Session History", footer credit. Status messages and error popups remain English for now.
• Your UI-language choice is remembered across launches.

Quick language switch (transcription)

• New "Switch to 中文 / English" button next to the Transcription Language dropdown — one-click toggle between EN and ZH for future recordings.

Re-detect last recording

• New "Re-detect in 中文 / English" button: if you spoke in the wrong language, click it to re-run Whisper on your last recording in the other language. No need to record again.
• Works for batch ("Record then transcribe") mode. The saved audio lives at %LOCALAPPDATA%\VoiceTyper\last_recording.wav (one file, overwritten each session).
• New result lands in the Latest Transcript box and auto-pastes into your active app, just like a fresh recording.

How to install

1. Quit any running VoiceTyper (right-click tray mic → Quit).
2. Download VoiceTyper-Setup-v1.2.0.exe below.
3. Run the installer — it detects your existing v1.1.0 / v1.0.0 install and upgrades in place. Your hotkey, mode, and language settings are preserved.

Voice Typer v1.1.0

What's new in v1.1.0

Live dictation mode

• New Live dictation option (in addition to the original "Record then transcribe" mode). Press the hotkey, start speaking, and text appears in your target app phrase by phrase as you pause — no need to press stop first.
• Voice-activity detection waits for a ~1-second natural pause to commit each sentence. That gives Whisper a full sentence of context per call, so accuracy is much closer to batch mode than the usual "transcribe every 500 ms" streaming approach.
• Mode switcher in the main window; your choice is remembered across launches.

Floating "Listening" overlay

• A small always-on-top indicator appears at the top-right of your screen while recording, showing a red dot + "Listening..." with animated dots.
• Works in both live and batch modes. Doesn't steal focus from your target app.
• Language badge on the right — "· EN" or "· 中" — tells you which language the model is set to at a glance.

Accuracy improvements

• Previously-transcribed text is fed to Whisper as initial_prompt context, which helps with homophones, proper nouns, and mid-sentence continuation.
• beam_size bumped to 3 for better short-phrase word choice.
• Ellipsis post-processing — Whisper's tendency to sprinkle ... at phrase boundaries is stripped out.
• VAD threshold lowered so quieter speech / further-from-mic still registers.

Fixes

• Missing pystray and Pillow added to requirements.txt (were imported but not listed).
• Single-instance mutex is more robust across launch methods (exe vs. python main.py vs. setup_and_run.ps1).

How to install

1. Quit any running VoiceTyper (right-click tray mic → Quit).
2. Download VoiceTyper-Setup-v1.1.0.exe below.
3. Run the installer — it detects your v1.0.0 install and upgrades in place. Your hotkey / mode / language settings are preserved.

Voice Typer v1.0.0

Voice Typer v1.0.0

First release of Voice Typer ??a speech-to-text desktop app for Windows.

How to install

1. Download VoiceTyper-Setup-v1.0.0.exe below
2. Run the installer
3. Follow the setup wizard