VoiceDumps 0.2.0

Reading a transcript and following one are different jobs, and until now the app only knew how to do the second.

Reading

Reading and following are now separate modes. Open a note with the player idle and you get an ordinary page: one size, full contrast, nothing chasing you. Press play and it follows the spoken word, growing the current paragraph and pulling the rest back — the behaviour that was always there, now only when it makes sense.

Timestamps hang in the margin and seek when you click them. Same measurements the PDF export already used, so a printed transcript and the screen finally share an anatomy. On a narrow window the stamp moves above its paragraph instead of squeezing the line.

⌘F finds within the open transcript — match count, next and previous, highlighted in place.

Keyboard: Space plays and pauses, ←/→ skip five seconds, ↑/↓ step by paragraph. The measure is a little narrower, closer to the 66 characters a line is comfortable to read.

Dictation

The first dictation after launch is faster. The model now loads while you are still speaking rather than after you release the globe key, which was the one moment the delay was impossible to hide.

Install

Download the DMG, drag VoiceDumps to Applications, open it. macOS will refuse the first launch because the app is signed but not with a paid Apple certificate — open System Settings → Privacy & Security and click Open Anyway.

Two permissions: Microphone, prompted on first recording, and Accessibility for the globe key, which has to be granted by hand in Privacy & Security → Accessibility. No app can grant that automatically.

Apple Silicon only. Everything runs on your Mac — no account, no API key, no network calls.

sha256  26591222e92eb9ffefa801a7716f147b046707e3b027dfa7f4d1c775524147c3

VoiceDumps 0.1.0

Local speech to text for macOS. Dictate anywhere with the globe key, record from
the microphone, or drop in an audio or video file and read the transcript back.

The speech models ship inside the app — nothing is downloaded on first run,
there is no account, and it works with the network off.

Install

1. Open the DMG and drag VoiceDumps to Applications.
2. Open it. macOS will refuse, saying it cannot verify the developer — the app
   is signed, but not with a paid Apple Developer certificate.
3. System Settings → Privacy & Security, scroll down, Open Anyway.

Then two permissions:

• Microphone — prompted the first time you record.
• Accessibility — System Settings → Privacy & Security → Accessibility.
  Required for the globe key: watching for a modifier key means installing an
  event tap, and macOS never grants that automatically.

Because the signature is ad-hoc, each new release is a different app as far as
macOS is concerned, so Accessibility has to be granted again after an update.

Apple Silicon only. Intel Macs need a rebuild from source.

In this release

• Globe-key dictation, with a floating level indicator
• Microphone recording, drag-and-drop files (mp3, m4a, wav, flac, ogg, mp4, mov)
• A reader with word-level playback follow-along and inline editing
• Export to PDF (typeset, not a screenshot), Markdown, or plain text
• Full-text search, light and dark themes, menu-bar background operation
• Picks the medium model (514 MB) on Macs with 16 GB or more, small
  (181 MB) below that. Override with VOICEDUMPS_MODEL_SIZE=small|medium.

Verify the download

shasum -a 256 VoiceDumps_0.1.0_aarch64.dmg
2869123c7c45507f884d48de96a8b6fc440f177576382f1d99ec6f06ebbc0914

Built from the source in this tag.