Fix garbled subtitle characters on Mac

Recover the text before saving a new encoding

If a subtitle shows café instead of café, first reopen an untouched copy using the encoding that originally produced the file. Once the words display correctly, save a separate plain-text UTF-8 copy and check it in the destination player. Saving already-garbled text as UTF-8 can preserve the mistake rather than repair it.

An encoding is the rule used to turn stored bytes into characters. The file extension identifies a subtitle format; it does not prove which encoding the bytes use. Work with subtitles you own or have permission to edit and distribute, and keep the original file until the repaired version passes review. This workflow does not require uploading the text to an online converter.

Separate three symptoms before changing anything

Repeated extra characters around accented letters, such as café, suggest text decoded with the wrong character encoding. This kind of garbling is often called mojibake. It is a clue, not enough evidence to identify the original encoding: use the exporter’s settings or ask the person who supplied the file.

A box where a character should appear can instead be a font or player rendering problem. Compare the same text in a Unicode-capable editor and another player before rewriting the file. If the editor shows the intended character and only one player shows a box, investigate that player’s font support and settings.

A replacement symbol, often displayed as �, needs more caution. Unicode documents it as one possible response to invalid encoded data. If that symbol has already been saved into the only surviving copy, changing encodings cannot reveal which original character it replaced. Seek an earlier file or a fresh export, or correct the passage against an authoritative transcript. Wrong timing and wrong translation are separate problems; an encoding conversion will not fix either.

Keep an untouched original and identify the source

Duplicate the file in Finder before opening it for repair. Use distinct names such as workshop-original.srt and workshop-utf8-review.srt. Keep the original recording and the original subtitle export together so that a readable but incorrect name can still be checked against its source.

Ask which program exported the subtitle and which text encoding was selected. For example, Windows-1252 may be relevant to an older Western European file, but it is not a universal Windows or European-language default. A filename, language or operating system alone is not a reliable encoding detector. If the source is unknown, try candidate encodings only on the duplicate, without saving over the original, and inspect several known words across the file.

Reopen the copy with a deliberate encoding

Apple documents a per-file choice in TextEdit: choose File > Open, select the file without opening it, then click Show Options. In Plain Text Encoding, choose the known source encoding and open the file. If the option you need is missing, Apple provides Customize Encodings List. These are TextEdit controls, separate from Polyglot’s subtitle import.

Inspect accented names, punctuation and words from every script present. A readable first sentence is not enough if a later speaker’s name is still damaged. If a candidate makes the text worse, close the working document without saving and reopen the untouched copy with another supported choice.

If the same garbled letters remain under the confirmed source encoding, the damage may already be part of the saved text. Return to an earlier export instead of repeatedly saving and reopening the damaged version. Avoid a global replacement of a suspicious character sequence: some occurrences may be legitimate, and different corruption paths need different corrections.

Save readable text as UTF-8, keeping the subtitle syntax

Once the text is correct, save a new plain-text file using UTF-8. Apple’s TextEdit settings include separate opening and saving choices under Plain Text File Encoding in Open and Save. Record any default you change and restore it afterward if other documents depend on it. Keep the repaired subtitle as a separate copy and reopen that copy as UTF-8 to confirm the result.

Check that the saved filename ends in .srt or .vtt as appropriate, rather than an added .txt extension. Do not save rich text and merely rename it .srt. Disable smart punctuation or text replacement while editing subtitle syntax: replacing the two hyphens in a timing arrow can break the cue.

W3C specifies UTF-8 for WebVTT. SRT and WebVTT also have different file structures, so changing .srt to .vtt is not a format conversion. Likewise, changing a label to UTF-8 does not convert the stored bytes. Use an actual save/export encoding option, and handle any format conversion as a separate step.

Check a complete cue, not just the accent

This fictional SRT cue illustrates a text-only repair. If the intended sentence is confirmed as “Meet Zoë at the café.”, the reviewed cue below keeps its number and both timestamps. The sentence is original sample content, not a model-accuracy test or a transcript from a customer.

Compare cue count, identifiers, order, start and end times, blank separators and intentional text line breaks with the source. Correcting characters should not silently renumber cues, shift a scene or remove an overlapping caption. Review every affected word; not every visible change is automatically a correction.

17
00:00:42,200 --> 00:00:45,600
Meet Zoë at the café.

What Polyglot can accept after recovery

Polyglot’s SRT importer supports UTF-8, with or without a byte-order mark, and UTF-16 when a byte-order mark identifies its byte order. It does not automatically guess legacy encodings such as Windows-1252. Its WebVTT importer expects UTF-8. Recover a legacy file in a suitable text editor before importing it; an import error is not a reason to overwrite the source.

Polyglot exports subtitle text as UTF-8 and provides local editing and separate SRT/VTT files. It preserves imported cue timing and order, while WebVTT presentation details are flattened with a notice. Keep an archival original when styling or positioning matters. Successful import does not prove the words are correct: mojibake can itself be valid Unicode text, and translation cannot reliably reconstruct missing characters.

Verify the delivered file in its real destination

Open the exported subtitle beside the matching final video in the player or publishing system that will deliver it. Inspect the first affected name, a passage in the middle and the last affected cue, then review the complete repaired track. Check accents, punctuation, different scripts, line breaks and the language-track label. Confirm that the text still appears at the right moment.

If the local UTF-8 file is readable but the uploaded version is not, download the delivered track when the platform allows it and compare the same cue. Check the platform’s supported format and encoding instructions before another conversion. Keep the original, reviewed file and final delivered version separately. Stop when the available source cannot establish the intended characters; a fresh export or a verified correction is more useful than a plausible guess.

Keep exploring.