Dictate prompts and scripts with the microphone button

Describing a sound, a scene or a mood takes more words than most people want to type. Saying it out loud is quicker, and the phrasing often comes out more natural. DepCut puts a Dictate button on its prompt boxes, so you can speak a description and edit the text afterward.

Where the microphone appears

A small microphone button sits in the corner of these boxes:

• Describe the music and the voiceover Script in the Audio tab.
• The sound description box in the Sound FX tab.
• The prompt boxes in the Image and Video generation tabs.
• The assistant's chat box.

Hover the button and its tooltip reads Dictate.

Dictate in four steps

1. Click the microphone button. The first time, your browser asks for permission to use the microphone. Choose Allow.
2. Start speaking. The box is replaced by a panel that shows your words as they come, over a level meter. It reads "Listening…" until the first words arrive.
3. Press the check mark (Use transcription) to accept, or press Enter. Press the cross (Cancel) or Escape to throw the take away.
4. The recognized text is added after whatever you had already typed in the box. Edit it like any other text.

If the browser has blocked the microphone, DepCut shows a short message asking you to allow the microphone for the site. Change the permission in your browser's site settings, then try again.

With the Mac app and without it

With the DepCut Mac app running, dictation is live. The microphone audio goes to the app on your Mac, speech recognition runs on the device, and the words appear while you talk.

Without the app, the browser records your take and sends it for transcription when you press the check mark. You see "Listening…" while you speak, and the text arrives a moment after you confirm. The result is the same text in the same box.

Speak a prompt that works

A dictated prompt is still a prompt, so the usual habits apply.

• Start with the subject, then add the mood and the details. "A slow piano piece, soft and hopeful, with room for a voice on top" gives a music model plenty to work with.
• Say complete phrases. Short fragments leave gaps the model has to fill by guessing.
• Pause before you press the check mark and read what appeared. Fix a misheard word right in the box.
• Dictate a first draft, then type small corrections. Mixing the two is quicker than doing either alone.

Because the new text is added after what is already in the box, you can build a prompt in pieces. Dictate the subject, confirm, then dictate the style as a second take.

When dictation helps most

Dictation fits any place where you describe something: a sound effect, a track, a shot, a request to the assistant. It also suits a voiceover script you want to draft in your own rhythm. Pick an AI voice afterward and let it read the cleaned-up version.

Generation boxes still spend credits when you press their Generate button. Dictation is free when the Mac app is running; without it, the hosted transcription of your take spends credits.

Try it

Open DepCut, go to the Audio tab, press Dictate in the music box and describe the track you hear in your head.