How to Use Text to Speech on CapCut (App, Desktop, Web)

Text to speech in CapCut turns typed words into a spoken voiceover using built-in AI voices. It's handy when you don't want to record your own voice, when you're in a noisy office, or when you need a quick narration for a product demo. The trick is that it only works on a text layer, so you always start by adding text.

Text to speech in the CapCut app

1. Open the CapCut app and create a project with your video.
2. Tap Text and add text to your video.
3. Type your script and confirm.
4. Tap the text layer on the timeline to select it.
5. Choose Text-to-Speech from the bottom menu.
6. Pick a voice style and tap Generate.

CapCut creates the voiceover as audio on your timeline. Play it back to check pronunciation before you move on, especially for names, places and any industry terms your customers use.

Text to speech on CapCut desktop

1. Open your project in the CapCut desktop editor.
2. Add a text layer to the timeline.
3. Type your script into the text box.
4. Select the text clip.
5. Click the Text-to-Speech option.
6. Choose a voice, generate the audio and preview it.

Text to speech in CapCut Online

1. Open your project in the CapCut web editor.
2. Click Text and add a text box to the timeline, then type your script.
3. Select the text layer and click Text-to-Speech in the toolbar.
4. Choose a voice and language, then click Generate.

Where to find it on each version

CapCut version	Where Text-to-Speech is	What you pick
Phone app	Bottom menu after tapping the text layer	Voice style
Desktop	Text-to-Speech option with the text clip selected	Voice
Online	Toolbar, with the text layer selected	Voice and language

Why the voice won't generate

CapCut's help page lists five reasons text to speech fails. Most of them are quick to fix:

• No text selected. TTS only works on text layers. If you haven't selected one, it won't generate audio.
• Weak connection. CapCut says it relies on cloud processing, so a patchy connection can interrupt it.
• Unsupported voice or language. If the voice or language you picked isn't supported, it may fail. Try another.
• An old version or a glitch. Update the app or refresh the browser.
• It's there but silent. Sometimes the voice is generated but muted, or placed outside the timeline range. Check the audio track and volume.

That last one fools people the most. You tap Generate, nothing seems to happen, and you assume it's broken, when the audio was created further along the timeline or on a muted track. Scroll the timeline and check before you try again.

Writing a script that sounds natural

AI voices read exactly what you type, so the script does all the work. A few habits help:

• Write how you speak: short sentences, everyday words.
• Use one text layer per sentence or idea, so you can time each line against the footage.
• Spell out anything awkward, like brand names, the way they're pronounced.
• Use commas and full stops to create pauses where you'd naturally breathe.

Splitting your script into several text layers also makes edits painless. If one line needs changing, you regenerate just that line instead of the whole voiceover.

Timing the AI voice against your footage

Once the audio is generated, it sits on the timeline like any other sound clip. Drag it so each line starts as the matching shot appears, and trim the video if the narration runs shorter than the clip. If a line runs too long for the shot, shorten the text and generate it again rather than speeding the audio up. It's also worth turning the original video's sound down so the AI voice comes through clearly.

When you've got several text layers, you can choose whether to keep the words on screen as captions or delete the visible text once the voice is generated. For accessibility I'd usually keep short captions on screen, because plenty of people watch social video with the sound off, especially on a commute or late at night.

Text to speech versus your own voice

For quick explainers, product tours and social clips, an AI voice is fine and saves time. For anything that's about you or your business story, I'd record your own voice. People buy from people, and your voice carries warmth no generated narrator can match. CapCut lets you do both, and my guide to freeze frames in CapCut shows another simple trick for explainer videos.

If you're scripting with AI too, it's worth having a play with other tools; my piece on how to use Claude Motion covers animated explainers.

My take

Add text, select it, choose Text-to-Speech, pick a voice and generate. If nothing happens, check you've selected a text layer, your connection's steady and the new audio isn't hiding muted further along the timeline. Write the script the way you'd say it, one line per layer, and it'll sound far less robotic.

Frequently asked questions

Where is text to speech in CapCut?

Select a text layer first. In the app, choose Text-to-Speech from the bottom menu; on desktop and online, use the Text-to-Speech option or toolbar button.

Why is CapCut text to speech not working?

CapCut lists a weak connection, an unsupported voice or language, no text selected, an outdated version, or audio that's muted or outside the timeline.

Does CapCut text to speech need the internet?

CapCut says it relies on cloud processing, so a weak or unstable connection can interrupt it.

Can I choose the language in CapCut text to speech?

In CapCut Online you choose a voice and language before you click Generate.

Can I use text to speech without adding text?

No. CapCut says TTS only works on text layers.