How Subtitle Preflight checks subtitle files

Method in one paragraph

Subtitle Preflight first parses the supplied SRT file or declared WebVTT subset, then keeps parser diagnostics separate from deterministic rule findings. Each parsed cue is checked against the visible named profile for timing, duration, reading speed, line limits, empty text, and adjacent duplicates. Fresh results can be exported locally with rule, cue, observed-value, threshold, profile, and limitation fields. The method checks only the supplied subtitle text; it does not compare subtitles with source media or grant certification.

Parsing SRT and WebVTT

SRT parsing reads blank-line-delimited blocks, optional numeric indices, comma-millisecond timing, and cue text. The declared WebVTT subset reads the header, identifiers, dot-millisecond timing, cue text, NOTE blocks, and common settings.

Technical rules and finding IDs

Cue IDs use deterministic parsed order such as cue-0001. Finding IDs combine the accepted rule ID, cue ID, and deterministic ordinal. Findings are ordered by cue and accepted catalog order, so the same supported input and profile produce the same IDs, values, severities, and summary.

• SRT_INDEX_MISSING_OR_OUT_OF_SEQUENCE
• TIMESTAMP_SYNTAX_INVALID
• TIMESTAMP_END_NOT_AFTER_START
• CUE_OVERLAP
• GAP_BELOW_PROFILE_MIN
• DURATION_BELOW_PROFILE_MIN
• DURATION_ABOVE_PROFILE_MAX
• CPS_ABOVE_PROFILE_MAX
• WPM_ABOVE_PROFILE_MAX
• LINE_COUNT_ABOVE_PROFILE_MAX
• CHARS_PER_LINE_ABOVE_PROFILE_MAX
• CUE_TEXT_EMPTY
• ADJACENT_TEXT_DUPLICATE

Reading-speed and line formulas

CPS counts visible non-whitespace Unicode characters after markup removal, divided by cue duration in seconds, and displays one decimal place. WPM counts whitespace-separated tokens divided by cue duration in minutes. Line count uses supplied text lines; characters per line remove markup but retain visible spaces.

Format sources and product rules

WebVTT syntax statements are informed by the W3C Candidate Recommendation Draft, WebVTT: The Web Video Text Tracks Format, published 20 May 2026. The implementation supports only the declared subset documented on this site and is not W3C-approved. SRT parsing behavior and all profile thresholds are documented product rules; the default thresholds are not presented as an external standard.

Read the W3C WebVTT Candidate Recommendation Draft dated 20 May 2026

Review the declared WebVTT subset

Profiles, sources, and user edits

The default is General readability v1.0, source Subtitle Preflight product default (not an external standard), reviewed July 23, 2026. It is a product default, not a universal standard. Editing any numeric value marks current results stale; resetting restores the accepted contract after confirmation.

Reproducible findings and reports

CSV and HTML generation use the same schema-shaped analysis result. Exports include generation time, safe metadata without an original path, the complete profile snapshot, all summary counts, parser diagnostics, every finding field, limitations, and the browser-local disclosure.

Manual review that remains necessary

Audio accuracy, caption completeness, speaker identification, meaningful sound coverage, and visual placement are explicit manual-review limitations.

Methodology change log

Version 1.0.0 · July 24, 2026: Browser-local Stage 07 parser, accepted P0 catalog mappings, contract-shaped results, cooperative cancellation, stale-result suppression, and local report exports.

Read the privacy boundary or run the local checker.

Run the documented checks

Use the browser-local checker to inspect a supplied SRT or WebVTT file under the visible profile. Review the limitations before using a report in a delivery decision.