Home / Guides / Speech bubbles
How to Add Speech Bubbles to a Video
A speech bubble does two jobs a plain subtitle cannot: it shows which line belongs to which person, and it gives the video a comic, meme-ready look. The catch is that a bubble has to stay attached to its speaker while both of them move. Here are the three ways to do it, from free and slow to fast.
Method 1: Text stickers in TikTok or CapCut
Both apps have a text tool with a bubble-style background. Add a text element, type the line, pick a filled background, and drag it near the speaker's head. Set the element's start and end time to match the sentence.
- Best for: one or two punchlines, or a static scene where nobody moves much.
- Watch out: each bubble is a separate element you place and time by hand. A dialogue video with ten lines becomes ten timed elements, and if the speaker walks across the frame, the bubble stays behind.
Method 2: Keyframing in a video editor
In CapCut desktop or any real editor, you animate the bubble's position with keyframes so it follows the speaker. Add a position keyframe when the head moves, drag the bubble, add another keyframe, repeat.
- Best for: a polished piece where the bubble has to track the speaker exactly, and you have the patience.
- Watch out: this is the most time-consuming option by far, and one more line of dialogue means one more animated element. It is the right tool for a hero video, a poor fit for daily posting.
Method 3: Automatic attachment
Tools like Vinnipuff detect the speech, build the bubble from the words, and attach it to the person who is talking, keeping it anchored as they move. You fix the occasional misheard word and pick a style; the tracking is not your job.
- Best for: dialogue videos you post regularly, where hand-placing every bubble is not sustainable.
- Watch out: automatic speaker detection can hesitate when two people talk over each other. Choose a tool that lets you reassign a line to the right person in a click or two.
Five rules for readable bubbles
- Tail to the mouth. The bubble's tail points at the speaker's face, not their body. Viewers read the tail subconsciously as attribution.
- One idea per bubble. A bubble holds one sentence, two at most. A paragraph inside a bubble stops being a bubble.
- Border it. A bubble without a dark border melts into bright backgrounds. A 2px outline and a small shadow keep it readable over anything.
- Stay out of the UI. Top and right edges of vertical video get covered by platform buttons. Keep bubbles in the middle band of the frame.
- Pop, do not fade. Bubbles are a comic device. A quick scale-in reads as intentional; a slow fade reads as a mistake.
Vinnipuff is built around this exact format: speech bubbles that attach to the right speaker and follow them through the scene, styled in comic, chat and manga looks. In early access now.
Join early access