How do I add subtitles to a video automatically?

Upload the video to Valmera and type "add subtitles". The AI agent transcribes the speech with a timestamp for every word, groups it into readable lines, checks the timing, and burns the subtitles into the video. Fix any word in chat, then export a full-quality MP4.

How accurate is the subtitle timing?

Each line starts on its first spoken word and clears at the end of the thought, because timing comes from word-level timestamps rather than sentence estimates. Subtitles follow every cut automatically. Before replying, the agent audits the track for late lines, words left uncovered and overlapping lines.

How do I fix a wrong word in the subtitles?

Tell the agent: "at 1:42 'cash flow' should be 'cache flow'" fixes one occurrence; "change X to Y everywhere" fixes all of them. You can also correct the line in the transcript panel. Corrections change the text only, never the audio.

Can it make subtitles in another language?

Yes. Ask for subtitles in another language and the agent translates each transcript segment in full and burns it in at the original timing, in a readable subtitle style. Right-to-left scripts such as Arabic and Hebrew render correctly. Have a fluent speaker review the translation.

Which languages can it transcribe?

The spoken language is detected from the audio automatically. English is the best-tested language; for others, check names and technical terms before you export.

Can I export the subtitles as an SRT or VTT file?

No. Valmera burns subtitles into the picture, with no SRT or VTT import or export. That makes them visible on every platform and in every repost, but there is no separate file to upload for closed captions.

Is the subtitle generator free?

Creating an account and uploading are free. Editing, including subtitles, requires a subscription. Exports are full-resolution H.264 MP4s and end with a brief Valmera end card.

Auto Subtitle Generator with Accurate Timing

To add subtitles to a video automatically, upload it to Valmera and type “add subtitles.” The AI agent transcribes the speech with a timestamp for every word, groups it into readable lines that appear on the first spoken word, and burns them into the video so they show on every platform, with or without sound.

Fix any misheard name in one message. Ask for another language and the agent translates every line at the same timing.

Subtitle Your Video

Upload, type “add subtitles,” review the preview. Creating an account and uploading are free; editing requires a subscription.

Create an account →

See pricing →

How to Generate Subtitles in Valmera

1
Upload the video
MP4, MOV, M4V, MKV or WebM, up to 14 GB and 3 hours. Indexing transcribes the speech with word-level timestamps and detects the language.
2
Ask for subtitles
Type “add subtitles,” or set the format: “documentary style, one line at a time.” Add “in Spanish” for a translation.
3
Check names and numbers
Read the transcript panel for brand names, people and figures. Correct a line there, or tell the agent the fix.
4
Review and export
The agent audits timing and coverage and checks rendered frames for contrast. Watch the preview, then export a full-resolution MP4.

PROMPTS TO PASTE

Add subtitles so it can be watched on mute.

Subtitles in the documentary style, one line at a time, at least three words per line.

Add Spanish subtitles.

The subtitle at 1:42 says “cash flow”; it should be “cache flow”.

How the Timing Stays Accurate

Question	Answer
Where timing comes from	A start and end time for every word, not an estimate per sentence
When a line appears	On its first spoken word
When a line clears	At the end of the thought; lines reset at breaths and sentence ends instead of hanging through silence
After you cut	Every line follows the edit automatically, so cuts never knock subtitles out of sync
Before you see it	The agent audits the track for late first lines, uncovered words, overlaps and lines that wrap

The difference shows on fast speech. Subtitles built from sentence-level timing appear early or linger; subtitles built from word-level timestamps land when the words do. The captions docs show how timing maps through cuts.

Settings That Matter

Setting	Default	How to change it
Style	Chosen for the content	“Documentary style” for a translucent panel, “clean” for plain white phrases.
Words per line	Set by the style	“One line at a time,” “at least three words per line,” or a maximum from 1 to 16.
Size and position	Medium, bottom	“Smaller,” “XL,” “move them to the top.”
Contrast	Outline and shadow	“Add a dark backing panel” for bright or busy footage.
Language	Detected from the audio	“Add French subtitles” for a translation at the same timing.

For lessons, webinars and interviews, a full phrase per line on a translucent panel reads best. For vertical short-form, switch to a punchier caption style with one or two words at a time.

Common Mistakes

Skipping the name check. Speech recognition guesses at brand names, people and jargon. A two-minute read of the transcript catches most errors.
White text on a bright background. Add a backing panel or move the subtitles where the frame is darker.
Sentence-length lines on vertical video. They wrap into blocks that cover the picture. Use shorter groups.
Publishing a translation unreviewed. Machine translation is a strong draft; have a fluent speaker check it.

Limits

Subtitles are burned in, with no SRT or VTT import or export, so there is no file to upload as closed captions and viewers cannot turn them off. English is the best-tested transcription language. Overlapping speakers are harder to transcribe than one voice at a time. If you already have subtitle files, the free subtitle timing shifter and SRT to VTT converter work on them in your browser.

Related Tools

For styled short-form looks, see add captions to video and word-by-word karaoke captions. Clean the audio first with silence and filler-word removal; subtitles follow every cut. Workflows for course creators, webinars, podcasters and YouTubers show subtitles in context, and the AI video editor page covers everything else.

Over Valmera's MCP server, subtitles use add_captions, with set_caption_fixes for corrections and audit_captions to check timing.

Frequently Asked Questions

Made with Valmera: real exports

Generate Accurate Subtitles

Timed to every word, corrected in chat, burned into a full-quality export. Creating an account and uploading are free; editing requires a subscription.