Tuna

Talk Mode

Talk Mode is dictation.

More precisely, it is a fast way to get spoken words into Tuna so you can keep going from there.

Local First

Talk Mode uses local models on your machine.

That means your speech does not need to leave your Mac just to become text.

Cloud systems may win benchmarks. But I find for most of my use cases, the local models perform surprisingly well.

Recordings have no fixed duration limit. The elapsed timer appears as soon as recording starts.

Talk Mode stays in a dedicated floating window at the screen edge. By default it remains above other windows without taking keyboard focus from the app you were using. Enable Focus window when Talk Mode starts in Settings to use Return, Escape, and Tab immediately; Tuna returns focus to the previous app after pasting or canceling. Drag the window anywhere for the current recording; it returns to the position chosen in Settings next time. Press the Talk Mode hotkey to submit, or focus the dictation window and press Return. Press Escape to cancel. You can also choose whether the transcript is staged in Tuna or pasted into the app that was focused when recording began.

With a shortcut trigger, Tap or hold lets a quick tap start a recording that stays open until the next tap, while holding the shortcut records until you release it. The separate Hold to talk and Toggle behaviors remain available. Choosing Paste directly applies to every Talk Mode activation behavior, including recordings committed by releasing a held shortcut. Modifier-style triggers use the same activation behavior, including Tap or hold.

If you trigger Talk Mode again while transcription, post-processing, or pasting is still underway, Tuna keeps finishing the pending transcript. You can start the next recording after it lands.

While a transcript is being pasted, canceling Talk Mode stops any remaining typing. Text already pasted stays in the destination app. An empty response from an AI post-processing prompt reports an error instead of leaving the transcript waiting indefinitely.

What Happens Next

The transcript becomes text, which means it drops straight into Text Mode territory.

From there you can:

  • paste it
  • copy it
  • transform it
  • keep chaining actions

So Talk Mode is not really a separate universe. It is an input path into the same command structure as the rest of Tuna.

Input Monitoring

Double-press and side-specific modifier-hold triggers need macOS Input Monitoring permission. Tuna asks for it when you configure one and shows a warning beside the trigger if macOS still blocks the input listener.

When To Use It

Use Talk Mode when typing is the annoying part.

Finally, there’s Combo Mode.