Choose the spoken language and re-transcribe when needed

Automatic language detection remains the default. When you already know what is being spoken, you can choose that language before the job begins and give speech recognition a clear instruction from the start.

This is useful for short recordings, mixed-accent audio, proper-name-heavy interviews, and languages that can sound similar in a brief opening. The language list is searchable, so you do not have to scan a long menu.

If the result comes back in the wrong language, select Wrong language? beside the detected-language label. Pick the language you intended, review the confirmation, and Pepys creates a new transcription from the recording.

A re-transcription is charged like a normal job because the audio is processed again. Pepys shows that before it starts, rather than silently spending credits in the background.

Correcting the language creates a new job. It does not overwrite the first transcript, so you can compare the results or keep edits already made to the original.

Pepys can reuse an uploaded recording while its source media is still retained. If that recording has already been deleted under your retention setting, upload the file again to run it in another language.

The source-language control tells Pepys what language to recognize in the audio. Translation happens after transcription and produces another readable version while preserving the transcript's timing.

Choose the spoken language when recognition is wrong. Choose translation when the transcript is correct but the reader needs it in another language.