Get transcripts for original and dubbed audio
The transcript must describe the voice you are checking.
A translated subtitle file can be perfectly written while the recorded dub says something else. Comparing that file with the original script checks the planned text, not what the audience hears. Begin by identifying both recordings, then obtain a separate timed transcript for each.
Choose a path based on what you have
- Two audio or video files: transcribe each recording separately with a tool appropriate for the language and your data-sharing needs. Export timed captions.
- Captions from an editor: export the captions associated with each actual track. Check that the dubbed captions were generated from the recorded voice rather than copied from the translation script.
- A YouTube video you manage: confirm the intended original and dubbed language tracks in your publishing workflow. Do not assume an available subtitle track is a transcript of the dub. If you cannot obtain a verified transcript, work from your authorized production audio.
- Only a translated script: obtain the recording before treating the result as evidence about spoken errors. You can compare scripts, but label that review accordingly.
YouTube documents captions as text with display timing and supports basic UTF-8 SRT files. Its supported formats extend beyond this checker; export SRT or WebVTT rather than assuming every YouTube caption format can be imported here. See YouTube’s subtitle format reference.
Verify the transcript before importing
Listen to the opening, a passage in the middle and a passage near the end of each recording. Check the language, speaker, numeric wording and timestamps. If a transcript says 120 but its own audio says 240, record a transcription problem before attributing the mismatch to dubbing.
Keep timestamps relative to each recording. If you cut both files into a short clip, keep a note of the original timeline position so the reviewer can find it later. A constant start difference can be handled with the checker’s offset; gradually increasing drift needs transcript realignment or shorter clips.
Export and load the pair
Save UTF-8 SRT, WebVTT or timestamped JSON. This tool accepts up to 2 MB, 4,000 segments and 2,000 numeric occurrences per transcript. Put original speech on the left, actual dubbed speech on the right, and choose the English, Spanish or German numeric convention used in each file. See the format examples and import walkthrough.
For example, an original segment at 00:12 may correspond to a dub segment at 00:42. If that 30-second difference holds throughout the clip, use +30 seconds. If it changes from one passage to the next, do not force the whole video into that offset.
Keep the right expectations
The website does not download YouTube videos, transcribe recordings or access your channel. Imported text stays in the browser. A third-party transcription tool has its own handling of uploaded media, so select it separately. If you lack permission to process or share a recording, use the fictional practice pair instead.