MachinesFluent changed more in v1.1.2 and v1.1.3 than our blog reflected. We did not publish a standalone v1.1.2 product update, so this post covers both releases. The headline is not a microphone test or a collection of bug fixes. MachinesFluent can now transcribe existing audio and video files, and the app has a redesigned interface with a full theme system inspired by the theme-rich experience of OpenAI Codex.
MachinesFluent in its light theme, with Smart Dictation controls beside the resulting structured text.
What actually changed
| Release | What changed | Why it matters |
|---|---|---|
| v1.1.2 | File Transcription for queued audio and video files | Meetings, interviews, voice notes, podcasts, and recordings can become text without being replayed into a microphone. |
| v1.1.2 | A reworked interface with named dark and light themes plus custom colors and contrast | MachinesFluent can match the workspace instead of forcing everyone into one visual style. |
| v1.1.2 | Reorganized Settings and clearer model, support, and update pages | Important controls are easier to find and the app feels more coherent. |
| v1.1.3 | Theme selection during onboarding, a more useful Windows tray, and better multi-monitor dock behavior | The redesigned experience now starts on first launch and stays closer to the work. |
A new interface you can make your own
The Appearance page is no longer a small collection of disconnected color controls. v1.1.2 rebuilt it as a real theme system with dark and light modes, dozens of named presets, and direct control over the accent, background, foreground, contrast, borders, and sidebar treatment.
The direction was inspired by OpenAI Codex: a serious working tool can still give people meaningful control over how it looks. MachinesFluent includes its own dark and light identity alongside a Codex preset and a broad library of other visual directions. You can choose a complete preset in seconds or use one as the starting point for a custom interface.
MachinesFluent uses the same preset-first idea—dark or light mode, named themes, then detailed controls—while keeping its own interface.
This was more than a palette swap. Settings were reorganized around clearer destinations such as Home, Language Models, Speech Models, Smart Dictation, and Feedback & Support. Cards, controls, previews, spacing, and navigation were brought into one visual system so the application feels like one product instead of a collection of separate configuration pages.
v1.1.3 carries that work into first-run setup. New users can choose dark or light mode and select a named theme during onboarding, using the same controls found in Appearance later. The first experience now reflects the actual application instead of a temporary setup screen.
Turn recordings into text
File Transcription is the largest new capability in v1.1.2. Drop audio or video files into MachinesFluent, queue several at once, and turn them into text with the downloaded Whisper or Parakeet model already selected under Speech Models.
The File Transcription page accepts common audio and video formats, uses the selected local model, and lets you choose where transcripts are saved.
This opens a different workflow from live dictation. A meeting recording can become notes. An interview can become an editable transcript. A podcast, lecture, voice memo, screen recording, or video can become searchable text without playing the media back into a microphone or moving it into a separate transcription service.
The page supports drag and drop, an optional output folder, per-file progress, retry, and an immediate Abort control. Completed transcripts are saved beside the source file or in the chosen folder, while existing transcript files are preserved rather than overwritten.
Because this workflow uses downloaded Whisper or Parakeet models, the transcription stays on the machine. Model choice remains in Speech Models, so there is no second setup system to learn just for files.
A better fit for the Windows desktop
v1.1.3 makes the redesigned app easier to reach during ordinary work. The Windows tray can now show or hide the dock, switch the active input, open Hotkeys or Appearance, display the installed version, and show update status.
For multi-monitor setups, the collapsed dock can follow an ordinary click to the monitor currently in use when cursor following is enabled. There is still one dock; it simply stays closer to the active workspace. These are supporting improvements rather than the reason for the release, but they make the larger interface overhaul feel more natural across Windows.
Onboarding also uses the real microphone and theme controls from Settings. Input testing is useful there as a quick confirmation step, but it is part of a smoother setup experience—not the headline feature.
What about snippets?
Voice snippets are not new in these releases. They shipped in v1.1.0 and were covered in the v1.1.0 product update. A short spoken trigger can expand into a saved phrase or larger block of text locally after transcription. That remains a useful feature, but repeating it here would blur what v1.1.2 and v1.1.3 actually added.
Should you update?
Update now if you work with recorded audio or video, want the app to fit your visual preferences, or spend the day across several Windows applications and monitors. File Transcription and the new Appearance system are substantial additions, not maintenance details.
If you only use quick live dictation, the interface overhaul is still immediately visible. Choose a theme, open File Transcription with a short recording, and check the tray once. Those three actions show the real update faster than reading every technical release note.
Get the current version from MachinesFluent for Windows. The product now handles both sides of voice work: speaking live into any Windows app and turning recordings you already have into usable text.
Sources checked
- MachinesFluent public changelog — v1.1.2 and v1.1.3 release notes
- MachinesFluent homepage — current product context
- MachinesFluent download page
