Quiet Typist app

Changelog

Every release makes on-device transcription a little smoother. Catch up on new features, improvements and bug fixes here.

Download the latest version
Latest

v0.4.4Live transcription for free users, plus Lifetime Pro

New

  • Live transcription is now available on the free plan, up to 10 minutes per recording.
  • Added a Lifetime Pro purchase option.
  • The Account page now shows usage stats, with file transcriptions and live recordings counted separately.
  • You can now turn off automatic updates in Settings.

Improved

  • History now splits transcripts into paragraphs at pauses and sentence ends, and shows a speaker's name only once for consecutive lines.
  • After your first few recordings, you'll be reminded that you can import them again to identify speakers.
  • Failed transcriptions now show more specific reasons, such as a file that can't be decoded or no speech detected.
  • Redesigned the Membership page, with a toggle to switch billing periods.

Fixed

  • Fixed incorrect expiration dates for limited-time Max and Lifetime Pro.

v0.4.3Japanese and Korean interface, more stable transcription

New

  • Added Japanese and Korean interface languages.
  • When a transcription fails, you can now report the problem directly from the task list.

Improved

  • Organize with AI now handles longer English text, up to about 10,000 English words or 30,000 Chinese characters at a time.
  • Default speaker labels in Organize with AI now follow the interface language.
  • During long stretches of continuous speech, text is now split at pauses, so words are cut in half less often.

Fixed

  • If the Windows graphics driver fails, transcription now switches to the CPU with a notice instead of crashing.
  • Fixed missing spaces between joined segments in English, Korean, and other transcripts.
  • Fixed some features still being limited to the free plan for Max members.

v0.4.2Linux and Android apps

New

  • Added anonymous usage statistics, which you can turn off in Settings
  • Added a Linux app, with .deb packages for x86_64 and ARM64.
  • Added an Android app, with Alipay membership purchases and activation code redemption.
  • Recordings can now capture your microphone and computer audio at the same time (macOS 14.2 or later).
  • Added the lightweight multilingual Qwen3-ASR 0.6B model.
  • Added support for wma, amr, awb, aiff, aif, aifc, caf, m4b, and mka audio.
  • Added support for avi, flv, m4v, 3gp, ts, mts, m2ts, mpg, mpeg, and vob video.
  • Qwen models now let you choose the transcription language, with 30 languages supported.
  • You can now edit recognized text directly during live transcription.
  • You can now export logs from Settings > About and the Contact the author dialog to help report problems.

Improved

  • On first launch, Quiet Typist recommends Qwen3-ASR 0.6B or SenseVoice based on your computer's specs.
  • Speaker identification is more accurate, and short sentences are less likely to be assigned to the wrong speaker.
  • Better sentence segmentation, so long sentences are no longer broken into fragments.
  • The Membership page now shows how many devices each plan supports.
  • When selecting multiple items in History, you can now click anywhere on a row to select it.
  • Failed model downloads now explain why, and let you contact the author directly.
  • Going back to History from a transcript now returns you to the same page and position.
  • The desktop app now uses Quiet Typist AI by default and tells you clearly when a Max membership is required.
  • Model management now shows the hardware requirements for Qwen models.
  • Release notes are now easier to read.
  • You now get a clear message when you try to place an order with a revoked license.

Fixed

  • Fixed recognized text being lost when the transcript contained repeated content.
  • Fixed model downloads and sign-in failing behind corporate gateways or antivirus software that scans HTTPS.
  • Fixed memberships not being activated promptly after an Alipay payment confirmation timed out.
  • Fixed not being taken to the Membership page when a file shared from another app hit a membership limit.

v0.4.1Live transcription and dark mode

New

  • Added live transcription to the desktop app: text appears as you record and is saved to History when you stop (Pro).
  • Recordings can now use a microphone of your choice or capture your computer's audio (macOS 14.2 or later).
  • Recordings are saved to your Documents folder by default, and you can change the location in Settings.
  • Added dark mode, with Follow system, Light, and Dark options.

Improved

  • You'll now be notified if no sound is picked up after recording starts, or if recording permission is missing.
  • Transcripts up to 8,000 characters are now organized in one pass, and longer ones are split into sections by topic.
  • Organize with AI results are easier to read, and speaker colors match the original transcript.
  • You can now open the model storage folder directly.

Fixed

  • Fixed English words running together in transcripts.

v0.4.0Organize with AI now streams results

New

  • Added a Max plan, so you can choose between Pro and Max.
  • You can now choose the transcription language before transcribing, and your last choice is remembered.
  • Playback now has a speed menu, including slower speeds.
  • Added an MCP tool for saving transcripts to a file.

Improved

  • Organize with AI now shows results as they're generated, and you can expand it to see the reasoning.
  • For long transcripts, Organize with AI now drafts an outline before organizing each section, with headings that better fit the content.
  • The recommended VRAM for Qwen models is now 8 GB.
  • Transcribing long files now uses less memory.
  • Up to 512 tasks can now be queued; the free plan can queue one task at a time.

Fixed

  • Fixed a blank screen at startup and flickering when pages first opened.
  • Fixed sentences being cut in half when identifying speakers.
  • Fixed a license sync error appearing when your license had expired or you didn't have one.
  • Fixed overly long results being truncated but still saved when organizing with your own API key.

v0.3.2Fixed Windows crash on launch

Improved

  • The Tasks entry no longer appears in the sidebar when there are no transcription tasks.

Fixed

  • Fixed the app crashing on launch on some Windows PCs.
  • Fixed being unable to drag the window with touch or a pen on Windows.

v0.3.1Playback and editing in transcripts

New

  • You can now play the original file from a transcript; click a timestamp to jump there and highlight the current paragraph.
  • You can now edit text, find and replace, and rename speakers in a transcript.
  • Added the SenseVoice model, which supports Chinese, English, Japanese, Korean, and Cantonese.
  • You can now drag in a folder for batch transcription, and opus, webm, and mov files are now supported.
  • You can now search History, and Organize with AI results now include an outline.

Improved

  • Interrupted model downloads now resume automatically.
  • You now need to sign in before transcribing.

Fixed

  • Fixed duplicate recognition, speakers being merged by mistake, and timing offsets in files with multiple audio tracks.

v0.3.0Organize transcripts with AI

New

  • Added Organize with AI, with side-by-side comparison with the original, copying, and Markdown export.
  • Use the official AI, or bring your own API key for DeepSeek, OpenAI, and other compatible services.

Fixed

  • Fixed speaker identification getting stuck, and made it faster.

v0.2.1Export and record management

New

  • You can now turn off speaker identification.
  • You can now export to TXT and Word, including several files at once.
  • You can now delete multiple transcriptions at once.

Improved

  • The free plan is limited to 40 minutes per file; batch transcription and longer files require Pro.

Fixed

  • Fixed updates failing to install automatically at startup.

v0.2.0Free plan changes

Improved

  • Changed what's included for batch transcription and long audio and video files.

v0.1.1Multilingual recognition and subtitle export

New

  • Added the multilingual Qwen3-ASR model.
  • You can now export SRT subtitles in batch.

Improved

  • Transcription progress is now clearer.
  • Transcription now uses less memory.
  • You can now select a model by clicking its card.

v0.1.0First public release

New

  • Batch transcription of local audio and video files.
  • Timestamps and speaker labels.
  • Copy text and export SRT subtitles.
  • Transcription through MCP.