SRT Subtitle Generator for Audio and Video

How it works: upload, review, and export

Choose a supported recording or capture audio in your browser, select the source language when that helps recognition, and submit the task. Sign-in happens in place: after authentication you return to this route, and WhisperWeb preserves the selected file when the browser allows it. If the browser cannot retain that local File object, the page asks you to select it again instead of claiming an upload was saved.

The task stays on this page through upload, queueing, processing, success, and recoverable errors. A task ID in the address lets the owner reopen the server-side result after a refresh. The same task also appears in My records with its SRT-generator source, so generating subtitles does not create a disconnected second history.

When transcription succeeds, the Subtitles tab opens first. Listen to the source, check each numbered cue, save deliberate edits, and download the saved SRT version. Repeated export or cue editing reuses the completed transcription rather than creating or charging for another base transcription.

• Cue numbers are emitted in order, starting at 1.
• Start and end times use HH:MM:SS,mmm with a comma before milliseconds.
• Saved Unicode cue text is used by the downloaded SRT file.

Review subtitle text and timing

An automatic transcript is raw material, not a finished caption track. Replay names, numbers, speaker changes, and fast passages, then adjust cue text or start and end seconds. WhisperWeb rejects negative, empty, reversed, or out-of-order cues rather than exporting a file that only looks valid in the browser.

Cue editing is intentionally separate from the free-form transcript editor. Changing prose does not silently retime captions; when the transcript and cue versions differ, the workspace marks the subtitles for review. Save the cue set before downloading so the file you receive matches the version shown in your account history.

See an SRT example

This short, original spoken sample and its two-cue subtitle file show the exact structure produced by the download: an integer sequence, a start-to-end time range, subtitle text, and a blank line before the next cue. The comma in 00:00:02,100 is part of SRT timing; WebVTT uses a different header and timestamp convention.

Download the sample and import it as a caption or subtitle file in your editor. Match it to a four-second practice clip on the editor timeline, confirm both cues appear in order, and inspect whether your editor interprets UTF-8 text correctly. For a real project, export from the reviewed task above so the timing stays tied to your uploaded recording.

• 1 · 00:00:00,000 → 00:00:02,100 · Welcome to the subtitle review.
• 2 · 00:00:02,400 → 00:00:04,800 · Check every cue before you publish.

Supported files and plan limits

The chooser lists the media extensions accepted by the shared transcription service, including common audio and video containers. Browser-reported type and duration are only an early check; the server makes the final decision and verifies the current plan, duration, file size, and credits. The tool shows your available balance before submission and links to current plans and usage limits rather than hard-coding an allowance that may change.

Processing is cloud based, and private media, task results, and downloads require the task owner. Do not upload confidential or regulated recordings unless your own consent, retention, and organizational requirements allow it. Generated subtitles can still contain recognition and timing errors, especially with noise, overlap, accents, music, or specialized terms, so review remains part of the workflow.