Skip to content

Repair a confirmed UTF-8 decoding mistake

Repair text whose UTF-8 bytes were once read as legacy characters, such as Maré Serena instead of Maré Serena. Choose the repair from evidence about the mistake, then test it on one copied file. The action cannot infer a language or recreate characters already replaced with ?.

Recognize the byte path

The intended é has UTF-8 bytes C3 A9. Reading those bytes as Western legacy characters can produce the two visible characters é. The repair recovers those byte values and decodes them as UTF-8 again.

That is why this case uses Unicode (UTF-8), codepage 65001: it names what the original bytes should have meant. It does not mean “save this tag using an arbitrary UTF-8 switch,” and it does not change the audio codec. A different original byte encoding needs its own diagnosis and codepage.

Try one affected field

  1. Keep an untouched original and select one copied practice file.
  2. Inspect Extended tags… and identify the affected field and values.
  3. Choose Actions → Actions (Quick)…, then Convert Codepage.
  4. Use these settings and choose Apply action.
  5. Compare the resulting title with a known correct spelling and read any operation error before selecting more files.
Setting Value
Field TITLE
Codepage Unicode (UTF-8), 65001
TITLE before Result using codepage 65001
Maré Serena Maré Serena
Mira – live Mira – live
Harbour ♥ Harbour ♥
North Pier North Pier
Maré Serena Conversion error
Mar� Serena Conversion error
Mar? Serena Mar? Serena

The correctly accented title is intentionally an error case for this particular byte repair. Its characters do not encode a valid UTF-8 byte sequence through this reconstruction. An error is a reason to stop and inspect, not to try every encoding until one result looks plausible.

The last row remains readable ASCII but does not regain the lost accent. A previous replacement with ? or � can mean that the original character information is already gone. Compare an original backup or another reliable source instead of accepting a guessed name as an encoding conversion.

Keep the repair narrowly targeted

The action processes every value of the chosen field. A field containing already-correct and incorrectly decoded text can refuse conversion; it is not a per-character language detector. For this action, unrepresentable characters or invalid byte sequences report an error rather than silently returning replacement glyphs.

Use TITLE first. _TAG includes all tag fields; _ALL also includes filenames, and _DIRECTORY targets a folder name. Those scopes can affect correct text and paths. They are unsuitable for an unconfirmed whole-library experiment.

The byte recovery accepts characters up to U+00FF directly and supported Windows-1252 glyphs such as the punctuation in the practice cases. It is not a universal inverse of every historical mis-decoding. More complicated multi-stage damage needs a separately tested diagnosis.

Separate text repair from storage settings

Correcting a value does not choose ID3v2.3 versus ID3v2.4, change a sidecar file's encoding, or prove that a target player supports its eventual native tag. Review tag write options and text encodings for those separate choices.

$ansi evaluates with the application's ANSI context; it is not a replacement for selecting the known original byte encoding in this action. When importing new text, choose the correct source encoding instead of deliberately loading bad text and repairing it afterward.

Check timing and avoid a second repair pass

The action writes immediately. Saving it in a group stores a workflow without running it; applying that group writes the selected files. Do not run the same repair again just because the spelling was corrected once—the corrected accented text can now fail this same conversion.

A warning from one action does not stop later actions/groups. Avoid appending case conversion or deletion until this repair has been inspected; those can make diagnosis harder or change other data after a repair warning. Use Edit → Undo … for a recorded write in the current desktop session, and retain the original file for recovery beyond it.