Accurate captions make videos easier to follow, search, translate, and watch without sound. This Descript captions guide explains how to correct a transcript, label speakers, control timing and line breaks, apply readable styling, and export subtitle files or burned-in captions for YouTube, TikTok, Instagram Reels, and Shorts.
Overview
Descript’s text-based editing workflow gives you a useful starting point for captions because the transcript is connected to the media. The important work is not simply generating words on screen. It is reviewing the transcript against the recording, making captions readable at normal playback speed, and choosing an export method that fits the destination.
Use captions when viewers may need spoken dialogue plus meaningful non-speech information, such as “[music]” or “[door closes].” Use subtitles when the primary goal is translating or displaying dialogue. In practice, creators often use the terms interchangeably, but the distinction helps you decide what information to include.
Descript’s interface and available controls can change, so treat the names below as workflow guidance rather than a fixed map of buttons. In your project, look for the transcript, caption or subtitle styling controls, and export settings. Before publishing, preview the actual rendered file or uploaded draft instead of relying only on the editor view.
A reliable workflow has five stages:
- Generate or import the transcript.
- Correct the transcript while listening to the source.
- Review speaker labels, timing, line breaks, and emphasis.
- Apply a consistent visual treatment.
- Export either a subtitle file or a video with captions burned in.
If you are also cleaning podcast audio, complete the edit before doing the final caption pass. Removing filler words, rearranging sentences, or changing clips afterward can make the transcript and timing stale. See the podcast audio editing and cleanup workflow for a related process.
Checklist by scenario
For every Descript caption project
- Start with the final cut. Finish major edits before treating the transcript as final. If you edit from a transcript, play through the completed sequence once more after the edits.
- Correct names and specialist terms. Check people, brands, places, product names, acronyms, numbers, and words that automated transcription commonly mishears. Use the waveform or video preview to confirm unclear phrases.
- Check punctuation for meaning. Add commas, periods, question marks, and quotation marks where they help viewers understand the speaker’s intent. Do not use punctuation to create artificial pauses that are not present in the performance.
- Confirm speaker labels. Assign the right speaker to every section, especially in interviews, panels, and overlapping conversations. Keep labels short and consistent. If names are not known, use neutral labels such as “Host” and “Guest.”
- Review timing. Captions should appear when the words are spoken and disappear when the thought is complete. Avoid leaving a caption on screen after the speaker has moved to a new idea.
- Break lines by meaning. Keep related words together and avoid splitting names, short phrases, or a verb from its object across lines. Two short lines are usually easier to scan than one long line.
- Use emphasis sparingly. Bold, color, or capitalization can highlight a key term, but applying emphasis to every important word makes nothing stand out. Preserve the speaker’s meaning without turning captions into decorative text.
For YouTube videos and podcasts
For long-form content, prioritize accuracy, searchability, and comfortable reading over animated caption effects. Review the opening, names, chapter transitions, quoted material, and any section where the speaker talks quickly. If you are delivering a subtitle file, export the supported caption format available in your workflow, commonly an SRT or VTT file, and inspect the file in a text editor or caption preview before uploading it.
Keep a clean version of the transcript separate from platform-specific styling. That master transcript can support show notes, translations, clips, and accessibility revisions. The Descript transcript repurposing workflow shows how the same reviewed text can support additional formats.
For TikTok, Instagram Reels, and YouTube Shorts
Short-form viewers often encounter videos in a vertical player, so captions must remain legible without covering the creator’s face, product demonstration, or important on-screen text. Use a vertical project layout when appropriate, keep text inside a central safe area, and preview the video on a phone-sized screen.
Burned-in captions are often the practical choice for short-form exports because the text is part of the video itself. Use a restrained style: high contrast, a readable weight, limited words per caption, and enough time for viewers to read before the next caption appears. Avoid placing text at the extreme top or bottom, where platform interface elements may compete with it.
For translated subtitles
Translation should begin with a corrected source transcript. Review names, idioms, jokes, technical vocabulary, and references that may not transfer directly. If Descript provides translation controls in your current plan or workspace, use them as a draft and review the result in the target language. A literal translation may be grammatically correct but still sound unnatural or change the intended tone.
Keep translated subtitle files organized by language and version. Do not overwrite the approved source transcript when creating a translation. This makes later corrections easier and reduces the risk of publishing a mismatched audio and subtitle version.
Export comparison
| Destination | Preferred caption approach | Key checks |
|---|---|---|
| YouTube long-form | Upload a subtitle file when you want editable captions; use burned-in text when the design is part of the video. | Transcript accuracy, speaker changes, timing, file encoding, and readability on desktop and mobile. |
| YouTube Shorts | Use burned-in captions for a controlled vertical design, or the platform’s caption workflow when you want editable text. | Vertical safe area, text size, rapid pacing, and overlap with interface elements. |
| TikTok | Use a burned-in caption version when visual consistency is important; confirm the final preview before posting. | Face and product visibility, bottom and side safe areas, contrast, and line length. |
| Instagram Reels | Use burned-in captions for a predictable appearance across previews and reposts. | Vertical framing, safe margins, cover text, and legibility in a small preview. |
| Translation or archive | Export a subtitle file with a clear language and version name. | Correct language, synchronization, special characters, speaker labels, and complete coverage. |
What to double-check
Before export, watch the video with sound reduced or muted. This reveals whether the captions carry enough meaning on their own and whether the timing is comfortable. Then watch it with sound on and compare every caption to the spoken words. These two passes catch different problems.
- Accuracy: Names, numbers, negations, contractions, and technical terms match the audio.
- Completeness: Important sound cues and relevant speaker changes are represented where needed.
- Timing: Captions do not flash too quickly, lag behind the voice, or remain after a scene or speaker changes.
- Readability: Font size, weight, contrast, background treatment, and line length work on a small screen.
- Composition: Captions do not cover faces, lower-third graphics, product details, subtitles in the source footage, or platform controls.
- Consistency: The same speaker, emphasis, punctuation, and brand treatment are handled the same way throughout.
- Export: The chosen file or rendered video contains the intended captions, with no missing characters, blank sections, or unexpected timing shifts.
Keep a caption style preset or project template if your workflow supports it. Record the decisions that matter: preferred font, contrast treatment, speaker colors, maximum line length, safe-area guidance, and whether each platform receives burned-in captions or a separate file. A simple style note prevents every editor from making a different choice.
Common mistakes
- Trusting automatic transcription without review. Automated text saves time, but it can misread names, accents, overlapping speech, and context-dependent words.
- Editing the transcript after styling captions. A late cut can change timing and line breaks. Make the final edit first, then perform a final caption pass.
- Using one export for every platform. A caption file may work well for an editable long-form upload, while vertical short-form content may need a different burned-in layout.
- Putting captions too close to the edges. Platform controls and previews can cover text. Leave a conservative margin and inspect the destination preview.
- Overdesigning captions. Fast animations, multiple colors, outlines, and oversized text can compete with the speaker and reduce comprehension.
- Ignoring non-speech audio. A viewer who cannot hear needs relevant context such as music changes, laughter, alarms, or an off-screen speaker.
- Skipping the final render check. Styling can shift during export. Watch the actual file, not only the timeline.
For a broader short-form workflow, pair this checklist with the guide to turning long videos into shareable social clips. Captions are most effective when the clip’s framing, pacing, and on-screen graphics were designed together.
When to revisit
Use this checklist before every upload, but schedule a deeper review when the underlying workflow changes. Revisit it when Descript changes its transcription, caption styling, translation, or export controls; when a platform changes its video layout or upload options; when your brand introduces a new typeface or color system; or when you begin publishing in another language or aspect ratio.
It is also worth reviewing captions after seasonal planning cycles, when you are preparing a batch of episodes or campaigns. Test one representative long-form video and one vertical clip before applying the process to the entire batch. Confirm that your presets, safe areas, file naming, and approval steps still work.
For your next video, finish the edit, correct the transcript, label speakers, preview captions with and without sound, and export the version required by each platform. Save the approved style and a clean transcript alongside the final media. That small archive turns captioning from a last-minute task into a repeatable creator workflow.