Subtitle a YouTube Tutorial and Keep Editable Captions

Use Add Subtitles to Video when you need supplied caption text rendered into an MP4. For a YouTube tutorial, decide separately whether the destination should also receive editable closed captions. This example follows a spreadsheet explainer whose menu labels, spoken instructions, and caption files must remain consistent.

Choose the tutorial deliverables before burning any words

Priya has recorded a three-minute-forty-second explainer about filtering a project spreadsheet. She needs a ninety-second excerpt that demonstrates how to show only blocked tasks, plus a clean master for the complete lesson. The excerpt will have visible subtitles for a delivery where the words should remain in the picture. The main lesson should retain a separate editable caption track. These are related outputs, not interchangeable ones. Decide which version you are preparing before rendering. Otherwise it is easy to hand over a subtitled MP4 and discover that the recipient expected a caption file whose wording could be corrected without encoding the video again.

Burned-in subtitles become pixels in the video image. They cannot be switched off as a separate caption layer, restyled by the viewer, or changed by editing an accompanying SRT. Separate closed captions preserve a different kind of control when the destination supports them. YouTube Studio provides a subtitle workflow for adding caption files and editing caption text and timing. That workflow remains outside this browser utility. Choosing YouTube here does not connect Studio, create a channel caption track, or upload anything. If editable CC is the only required outcome, prepare the track rather than making a permanent-text video simply because rendering is available.

For this lesson, Priya records the intended outputs in a small handoff note: clean full lesson, full-source caption master, clean excerpt, burned-in excerpt, and clip-relative used SRT. The clean excerpt is useful if its recipient prefers separate captions later. The used SRT belongs to that excerpt, not automatically to the complete lesson. Keep the names descriptive enough that another editor can tell the difference without opening every file. This early distinction prevents a maintenance problem: a correction to the word “Blocked” should have a clear destination, instead of being applied to one text file while the delivered picture still shows the old wording.

Make technical terminology match the screen and the speech

Review the transcript with the actual spreadsheet recording visible. Priya says “Open the Status filter” while pointing to the Status column, then selects the value Blocked. A track from a previous draft might say Progress or Waiting instead. Those words are not harmless synonyms when the viewer is looking for a specific control. Compare every quoted label with the screen and every explanatory sentence with the audio. If the narration itself names an outdated control, a subtitle should not quietly present a different instruction as if that were what Priya said. Correct the recording or provide an explicit explanation in the source lesson.

Keep distinctions that protect the learner from doing the wrong operation. In this example, “Filtering hides the other rows; it does not delete them” is the key reassurance. Dropping the second clause to shorten the caption would remove the explanation of what happens to the missing tasks. Preserve negatives, quantities, and the names of selected values. If there are keyboard commands, check modifier names and punctuation against the recorded demonstration rather than expanding them from memory. This is a manual review of an existing track. The tool does not recognize controls in the screen capture, verify spreadsheet behavior, or detect that a technical term has changed between versions.

Treat a subtitle as speech support, not a replacement for all the information in the screen recording. A long table of project names should remain visible in the source rather than being copied into a cue. Conversely, a pointer circling an unlabeled icon may need a clearer spoken description in the lesson. Priya changes “click that” to a recorded instruction naming the Status filter before finalizing the transcript. The distinction matters: readable captions can preserve a clear explanation, but they cannot automatically supply a missing one. Review the lesson from the perspective of someone who has not seen the spreadsheet before and needs to connect each instruction with a visible target.

Map the short excerpt without damaging the full lesson timeline

Priya selects source 01:00 through 02:30 for the ninety-second excerpt. Her caption master still uses the full three-minute-forty-second recording. A cue at source 01:07.400 therefore belongs 7.400 seconds into this excerpt when no offset is needed. Keep those two time bases distinct. The tool applies the chosen global offset to source cue times first, intersects them with the selected interval, and writes the retained used SRT relative to clip zero. Do not overwrite the full lesson master with that download. Its shorter timeline is useful precisely because it describes a different deliverable from the original recording.

Inspect any sentence crossing the cut. A cue beginning at 00:59.500 and ending at 01:02.000 is retained only from the excerpt opening to output 00:02.000. The timing is mathematically valid, but its first words may refer to something the viewer never sees. Priya moves the opening earlier to include the complete explanation or chooses a later complete instruction, then checks the resulting interval again. Do not assume clipping cue edges makes the spoken lesson self-contained. The final instruction also needs its visible result: a filtered table should remain on screen long enough for the viewer to recognize which rows are still shown.

This tool accepts sources up to five minutes and 150 MiB, but each selected clip must run for at least 0.25 seconds and no longer than 120 seconds. Priya’s complete recording fits the source limit; burning the entire three-minute-forty-second lesson in one output does not fit the interval limit. Use the utility for the defined excerpt, and keep the full lesson’s editable caption workflow separate. For a longer source beyond the input limits, prepare an appropriate local source beforehand. These constraints are not YouTube upload limits. Avoid dividing a lesson into arbitrary two-minute pieces solely to satisfy the renderer when the real delivery requires one clean video with a separate caption file.

Keep closed captions as an independent correction path

Imagine the spreadsheet template changes a week after delivery. The column is renamed Review status, and the tutorial needs a revised narration. With a clean recording and an editable caption master, the next version has a straightforward source pair to update. With only a burned-in MP4, the old words remain visible even if a separate caption file is corrected. There is no switch that removes those pixels. Keep the clean video for future changes, and decide whether the screen recording itself needs replacement. Editable captions reduce one part of the maintenance work; they do not make an outdated screen demonstration accurate by themselves.

Plan separate caption versions when the audience needs another language or revised terminology. This utility does not translate the track or turn its visible text into selectable closed captions. A translated separate track must be prepared and reviewed through its own workflow. If it is displayed over an MP4 already containing the original language, both sets of words may compete for the same area. That can make the spreadsheet harder to inspect rather than easier. For the main lesson, a clean video with the intended caption choices often provides more flexibility than a permanent English block combined with another layer added at playback.

Do not describe hard subtitles as equivalent to closed captions merely because both show words. The viewer’s control, future corrections, and destination behavior differ. A readable rendered track can still be useful for a specific excerpt, but it does not certify that all accessibility needs have been met. Check whether meaningful sound information and speaker identity are represented, whether the visual explanation is understandable, and what the chosen player offers. Priya reviews the eventual CC track in its destination separately from the burned-in proof. Approval of one does not establish approval of the other, because their wrapping, position, and viewer controls are not the same rendering system.

Protect the menu, cursor, and result from the subtitle block

Screen tutorials place useful information close to the edges: column headers above, sheet tabs below, and menus that open over the table. Review the filter menu while it is expanded, not just the tidy spreadsheet before the click. A bottom subtitle might cover the sheet tabs, while a top subtitle can obscure the Status header. In Priya’s selected edit, the important menu occupies the upper left and the lower rows are not used in the demonstration. She reserves a consistent lower area for subtitles. That is an editorial choice for this recording, not a universal layout rule for YouTube tutorials or every spreadsheet interface.

The subtitle style uses a system sans-serif font, optional bold, chosen color, background opacity, and one vertical placement selected from top, center, and bottom, plus a margin. There is no per-cue cursor avoidance or automatic recognition of the active menu. Font sizing uses the shorter output edge: select three through eight percent, in half-percent steps. At a 1280 by 720 output, three and a half percent is 25.2 pixels in the renderer. Judge that size against the actual tutorial words and the screen details they must accompany. Making the entire video smaller can make both the subtitles and the spreadsheet labels harder to inspect.

Keep instructions short enough to fit without hiding the filter result. No subtitle box may exceed eighty-four percent of frame width, and the renderer permits no more than three wrapped lines. Overflow is rejected, not truncated. A long technical identifier may wrap awkwardly even when the cue is below the input character limit. Revise the cue boundaries in the editable track while retaining the exact term. A darker rectangle can help separate lettering from gridlines, but excessive coverage can conceal the row values the lesson is explaining. Test the longest instruction, the expanded menu, and the final filtered view before settling on the treatment for the whole excerpt.

Resolve track errors without throwing away the editable master

Bring a basic UTF-8 SRT or WebVTT copy that the burner can parse. Its track ceiling is 64 KiB, with no more than 200 cues and 500 characters per cue. Each interval must be ordered and must not overlap another. A larger caption master may be valid for a different workflow while exceeding these limits here, so keep it intact and prepare a suitable excerpt track deliberately. Do not silently delete lines until the file loads. Establish which part of the lesson the working copy represents, and keep its time basis documented so it can be paired with the right source video.

If an individual cue has a typo or bad duration, use the subtitle editor and sync fixer to correct the existing track before rendering. A constant delay can be tested with this tool’s signed global offset, with permitted values between negative 300 and positive 300 seconds. A track that matches the opening but drifts after a removed explanation needs more than one shift. Check a spoken reference near the start and another near the end. Multiple local edits, different takes, or a changed playback rate can require targeted timing work rather than a global offset chosen to make one convenient sentence look right.

Rich subtitle information needs particular care. The burner rejects structured blocks named STYLE, REGION, or NOTE in WebVTT, and conversion of markup or positioning information to plain text can lose intended presentation. Review the reported losses and acknowledge the plain conversion when required. The tool does not preserve an imported caption layout merely because the dialogue survives. Speaker labels and useful sound descriptions can remain written text, but they need human review too. Retain the original rich file and make the compatible copy a separate asset. That separation lets the full lesson keep its appropriate editable caption source instead of inheriting every simplification needed for this short burned-in version.

Check the exported tutorial as both a picture and a recording

The browser must support the source’s actual video and audio codecs, not just recognize an MP4, MOV, or WebM extension. Rendering uses native WebCodecs support and produces H.264 MP4. Long-edge options are 640, 1280, and 1920 pixels, capped at 2,073,600 output pixels, with no upscaling or cropping. Output is limited to thirty frames per second and 64 MiB. Choose a size that preserves the menu lettering in the source; a subtitle-friendly font cannot rescue an unreadable screen capture underneath. If the source itself is too soft, return to the recording rather than using larger caption text to hide that problem.

Keep Priya’s narration unless a deliberately silent excerpt is required. Requested audio becomes AAC at 128 kbps and 48 kHz, and a failure should stop conversion rather than silently omit it. Explicit mute is a separate choice. Listen to the downloaded MP4, then follow the filter operation from beginning to end. The cue naming Blocked should coincide with the appropriate selection, and the reassurance about hidden rows should accompany the changed table. A successful file download does not establish that the spoken sequence is complete, that the right menu remains visible, or that the listener has enough time to connect the action with the explanation.

Scrub the preview to inspect the expanded menu and cue boundaries, then play normally to assess the pace. Preview and export share the subtitle renderer, but the downloaded file and eventual destination still deserve inspection. There is no guarantee for every physical device or player. If a separate CC track is also supplied, check it on the exact video version it accompanies and look for duplicated words over the permanent subtitles. Priya’s review note distinguishes source timing, excerpt timing, and the destination caption check. Those are different checks, and none should be replaced by a screenshot of one attractive frame from the tutorial.

Package the lesson so a later correction is possible

Build a delivery folder around clear roles. Keep the clean full lesson and original full-source captions as masters. Store the clean excerpt, burned-in excerpt, and normalized used SRT under the excerpt’s version name. That SRT includes only visible cues after offset and interval clipping, starting from the selected clip’s zero; it does not preserve every cue from the full lesson. If someone asks to correct an instruction at minute three, return to the full-source master rather than searching the ninety-second track for a line it never contained. A small filename distinction prevents a surprisingly large amount of avoidable confusion during revisions.

The settings JSON preserves subtitle text, visual style, and time choices, but not the video or its source filename. Keep an external source reference if you need to reproduce the export, and treat the JSON as transcript-bearing material when sharing it. Review text, CSV, and the PDF document settings and review information; the PDF is neither the production recording nor an accessible-caption certification. This browser session has no autosave, so obtain the needed downloads before leaving. Local media processing does not upload the private video, but normal site analytics remain. Avoid interpreting local conversion as a promise that the entire website has no network activity.

Publishing the lesson and adding native captions remain separate tasks. The preparation label does not validate YouTube acceptance, connect an account, or send a file into SMTasker. For wider account work, review the SMTasker plans and trial options. SMTasker uses a connected Android device with a Windows or macOS local client. Its supported YouTube actions include comments, likes, and subscriptions, not YouTube publishing. Keep the final handoff focused on the approved recording and its matching caption version. The useful outcome is not simply words visible on a video; it is a tutorial that can be understood now and corrected later without losing the clean source.

Questions about YouTube Tutorials and CC

Continue with another subtitle preparation guide

Prepare your own clip with the Add Subtitles to Video, or continue with a focused guide. Find more utilities in all free tools.