JVoice 0.3.2

JVoice 0.3.2 adds a speed test that measures how fast you talk.

New

• Speed test: open Speed test in the sidebar, pick how long you want to talk, and press Start test. JVoice shows how fast you speak in words per minute and characters per minute.
• Pick 30 seconds, 1, 2, 5, 10 or 30 minutes, 1 hour, a custom length up to 3 hours, or No limit and press Stop when you are done. You can stop any test early.
• Your speed updates every few seconds while you talk, along with the words JVoice heard.
• The speech model writes down what you say, then JVoice counts the words and characters itself. No chat AI is used.
• Characters are letters and numbers. Spaces and punctuation are not counted, since you do not say them.
• Results are saved on your computer under Past results, with your best marked. Copy result puts your score on the clipboard.
• A test uses your OpenRouter key and costs the same as dictating for that long. While a test runs, the dictation shortcut and updates wait until it ends.

Improved

• Check for updates is now in the sidebar, under Settings. It shows when JVoice is checking, downloading or ready to restart, so you do not have to open Settings.
• The sidebar no longer shows a status dot and "Ready". It only shows a message when something needs fixing, such as "Shortcut not working".
• When JVoice is up to date, Settings > Updates now says which release is the newest and when it last checked, so you can see the check happened.

JVoice 0.2.0

First public release of JVoice: hold a shortcut, speak, and your words are typed into whatever app you are using.

New

• Hold Ctrl + Win anywhere, talk, and let go. The text is pasted where your cursor is. You can pick another shortcut, or switch to press once to start and again to stop, in Settings > Shortcut.
• Speech to text uses ElevenLabs Scribe through your own OpenRouter key. The key is stored encrypted on your computer.
• AI cleanup has two styles. Light cleanup keeps your words and removes um and uh, fixes punctuation, and applies "no wait" corrections. Rewrite turns what you said into clear text.
• DeepSeek V4.1 Flash is the default cleanup model. Settings > AI cleanup > Model also lists DeepSeek V4 Flash, DeepSeek V4 Pro, Claude Haiku 5.5, Gemini 3.8 Flash, Gemini 3.1 Flash Lite, GPT-5.4 Mini, GPT-5.4 Nano, Qwen3.8 Flash, Mistral Small 4 and Kimi K2.6. "Other model" lets you search every OpenRouter model.
• Extra instructions let you add your own rules for the AI, such as British spelling. They save as you type and apply to your next dictation.
• Settings > Output > Length limit caps the text at 128, 280, 500 or 1024 characters, or leaves it unlimited.
• History keeps your dictations on this computer. "Show changes" marks what the AI added and removed. Recordings that failed to send can be retried.
• Dictionary fixes names and words the speech engine gets wrong.
• JVoice updates itself from GitHub. It checks every hour and when your PC wakes up. New versions download in the background and install when you are not dictating, then JVoice restarts on its own. You can also check by hand in Settings > Updates.
• After an update, a notification says what changed, and Settings > Updates lists the changes.
• Your clipboard is put back after each paste, and the dictated text stays out of Windows clipboard history.
• Text pasted into a terminal goes in as one line, so a new line never runs a command.
• If you switch windows before the text is ready, it is copied instead of pasted into the wrong app.
• Updates install only if they carry a valid JVoice signature.