Changelog

1.2.0 — early preview

Voice translation has its own shortcut and page. Ordinary dictation no longer translates. If you had a language chosen, set a translate shortcut to keep translating.
Shortcuts: “Hold to talk” and “Tap to start and stop” are two rows, each with its own switch. The Hold/Tap choice is gone.
More cloud speech services: OpenAI, Groq, Google Cloud, Microsoft Azure, AssemblyAI and ElevenLabs, plus any OpenAI-compatible service. Not yet tested against the real services.
More cloud text recognition: Microsoft Azure AI Vision and Mistral OCR. One key per company is shared across speech, text recognition and My AI models.
Quiet speech: local models read soft recordings with noise reduction when a steady room noise is close to the voice.
Vocabulary packs, text tidying, AI polish and “My AI models” for polishing and translation.
Not notarized; macOS 14 or later, Apple silicon or Intel.

One package for Apple silicon and Intel Macs; the minimum system is now macOS 14 (it was macOS 26).
Recording bar: see-through glass in light mode; an optional character style with short animations.
On-device text recognition models for screenshots (PP-OCR).
Not yet verified on a real Intel Mac, or on macOS 14 and 15.

The first public build. It is used every day on the maintainer’s Mac but has not been tested on many setups yet.

Voice input: hold a shortcut, speak, and the text is typed at your cursor.
Local recognition with SenseVoice, FireRedASR2 and Parakeet; optional cloud services with your own keys and consent.
Screenshots and text recognition (new, still being tested).
An optional local API for your own programs and hardware.
Not notarized, macOS 26 or later, Apple silicon only.

Every release is listed, with its package and checksum, on GitHub Releases. The full, detailed list is the repository’s CHANGELOG.md.