1. Choose the Figma artifact before the microphone
A Figma Make prompt, a Design text layer, a comment, an audio conversation, live captions, and a voice-enabled product prototype are different artifacts. They have different permissions, storage behavior, recipients, and review points.
Start with the destination. Use Figma Make voice-to-text for an AI generation prompt. Use Dictation in text-edit mode for interface copy. Use Dictation in the comment field for feedback. Use Figma audio when collaborators should hear one another live. Use captions to read that live conversation. Build real microphone behavior in the product code rather than assuming a visual prototype supplies it.
- Generate or refine a Make prototype: native voice-to-text prompting.
- Write editable canvas copy: Dictation in a Figma Design text layer.
- Leave editable feedback: Dictation in the comment field.
- Talk to collaborators: Figma audio conversation.
- Read live team speech: Figma Desktop captions or Chrome Live Caption.
2. Use Figma Make’s native voice-to-text for prompts
Figma’s current Make product page lists Voice-to-text prompting and describes dictating ideas to generate a prototype. This is the native first choice when speech should become an AI prompt rather than ordinary canvas copy.
Figma Make requires microphone access for voice-to-text prompting. Access also depends on the current plan, seat, AI-credit limits, and organization policy: Figma documents full access for Full seats on paid plans, trial access for other seats and plans, and admin controls that can disable AI features.
Review the visible prompt before submitting it. A spoken prompt can change a functional prototype or web app, so names, requirements, scope, and destructive instructions deserve the same check as a typed prompt.
3. Dictate editable copy inside a Figma Design text layer
Figma Design requires can-edit access to create or change text. Select the intended text layer, then double-click it or press Enter to edit its contents. Confirm that the insertion point is inside the text before starting the Dictation shortcut shown under System Settings → Keyboard.
Stop Dictation and review the words, line breaks, typography, and layout before leaving text-edit mode. Figma’s fixed-size text layers can wrap and overflow as content grows, so a correct transcript can still damage the visual hierarchy or push text outside its intended frame.
- Select the exact text layer and enter text-edit mode.
- Confirm a blinking insertion point before starting Dictation.
- Stop, proofread, and inspect wrapping and overflow before sharing.
4. Dictate comments without confusing comment mode and design mode
Figma supports comments for anyone with at least can-view access. In the browser or desktop app, press C or choose the comment tool, place the comment at the intended location, click in its text field, and use macOS Dictation.
Review the destination, mentions, and visible wording before posting. Figma says comment mode prevents changes to canvas objects; press Esc or choose another tool when you need to return to design editing. Dictating a comment does not modify the text layer underneath it.
5. Treat Figma audio as a live conversation—not text entry
Figma audio lets collaborators talk inside a Figma or FigJam file from the desktop app or a supported browser. It is not available on mobile or in presentation view, requires microphone permission, and each file can hold one active conversation at a time.
This path sends your voice to the other people in the conversation. Figma says it and the audio service hosting the feature do not record or monitor audio conversations. That statement belongs to Figma audio; it does not describe Apple Dictation, Figma Make prompting, a browser caption service, or a third-party transcription tool.
6. Closed captions are temporary accessibility text
Figma’s native closed captions are available in the desktop app. Figma directs browser users to Chrome Live Caption. The caption box shows a live transcript with speaker identity and lets you scroll back to the point when you joined the conversation.
Figma clears that transcript from the caption box when you leave. Rejoining starts with an empty history. Use captions to follow a live discussion, not as a saved meeting transcript, a design-text source of record, or proof that audio was inserted into the file.
7. A microphone in a prototype is not automatically functional
A Figma Design prototype can visually represent a microphone control, but the mockup alone does not capture audio, request permission, or perform speech recognition. Those behaviors belong in the implemented product.
Figma also draws a precise boundary inside Make: the current Make preview cannot request microphone or camera access for the app being previewed, while a published web app can request those permissions. This is separate from granting Figma Make microphone access for its own voice-to-text prompt control.
8. Troubleshoot the exact voice path that failed
If Figma Make voice-to-text is missing or silent, check Figma’s microphone permission, plan and seat access, AI-credit availability, and organization AI policy. If Figma audio is failing, check its separate microphone permission, browser or desktop-app version, audio device, network, file access, and organization audio policy.
If macOS Dictation does not enter Design text or a comment, test the same shortcut in TextEdit. A failure there points to the Mac’s Dictation setting, language, microphone input, Voice Control interaction, or current processing requirement. If TextEdit works, return to Figma, enter a plain editable field, confirm the insertion point, and use a reviewed clipboard fallback if the custom surface still rejects text.
9. Use IraVoice only for the local cross-app text requirement
IraVoice 0.7.1 is a separate option for macOS 26+ Apple-silicon Macs when one held-key workflow should supply editable text across supported Figma, email, note, and developer fields, or when speech recognition should run on the Mac. Models download on first use; the full trial needs no activation, and paid continuation needs one Gumroad key check.
IraVoice is not a Figma plugin and does not use Figma APIs, inspect the canvas, generate designs, operate Figma Make, join audio conversations, create captions, save transcripts, implement prototype microphone behavior, submit prompts, post comments, or publish a prototype. It targets a supported focused editable field exposed through macOS Accessibility and falls back to the clipboard when direct insertion is uncertain.