ResourcesFile Transcription

How to Translate an Audio File into a Bilingual Transcript on Mac

Translate a saved recording on Mac, check missing passages, and export a bilingual transcript with the original text, translation, and useful timestamps.

2026-10-24 · 8 min
An audio waveform flows into two aligned transcript columns with one passage highlighted.

Choose a transcript, live captions, or translated audio

These outputs solve different problems:

  • What you need: Read a translation while a recording plays; Suitable output: Live translated captions
  • What you need: Review, search, and share the original and translation; Suitable output: A bilingual transcript
  • What you need: Load translated text alongside a video; Suitable output: A timed SRT or VTT subtitle file
  • What you need: Hear the recording spoken in another language; Suitable output: A dubbing or voice translation workflow

Make the bilingual copy easy to review

This guide is about the second task. A bilingual transcript is a text document; it does not replace the voices in the recording. If your goal is a subtitle track, use the SRT and VTT guide.

For example, suppose a colleague sends a recorded project briefing in Japanese and you need an English reading copy. The useful result is not merely a page of English. It is a reviewed document where you can find the Japanese source, its English meaning, and the point in the recording where the statement occurred.

Prepare the recording and the language resources

Use the best available recording that you are allowed to process. Confirm that it plays, contains the expected speech, and includes the ending. Keep the original file in a stable location so you can return to uncertain passages.

A filename extension does not guarantee compatibility. M4A, MP3, WAV, and video containers can contain different encodings; the application must be able to decode the audio in your particular file. Try a representative recording before committing a large archive.

Recognition and translation need separate preparation. Choose the language actually being spoken for transcription. Choose the language you want to read for translation. A language appearing in the recognition list does not automatically establish that every translation pair is available.

Alison Workspace checks translation availability against the Apple language resources on your Mac. Download the requested resources before starting a long job. Once the required resources are installed, its supported recognition and translation run locally. The app requires macOS 15 or later and Apple silicon, M1 or newer; Intel Macs are not supported.

Transcribe the source before polishing the translation

Open the menu-bar menu and choose Transcribe Files…. Then:

  1. Use Add Files, or drag the recording into the file queue.
  2. Check the filename and displayed duration.
  3. Select the spoken language and prepare the requested recognition resources.
  4. For a review-first workflow, leave translation off initially and start transcription.
  5. Open the resulting record and check its completion status.

Choosing a translation target during transcription is another option. The review-first route is useful when names, technical language, or quotations matter: you can correct the source before asking for a translation.

Check how much audio was actually processed. The file-transcription allowance is separate from the live-caption allowance. A duration limit, cancellation, or processing failure can leave a partial record. Translating that record will not recover speech that was never transcribed.

If a long recording stops early, resolve the cause before treating the document as complete. A polished opening paragraph can make an unfinished transcript look more convincing than it is.

Review the words and the speaker assignments

Use playback to compare the transcript with the recording. Check a passage near the beginning, another in the middle, and the last processed passage, then spend more time on sections you will quote or act on.

Pay particular attention to:

  • Names, product terms, and abbreviations.
  • Numbers, currencies, dates, and units.
  • Negation and conditions, such as “do not ship” or “only after approval.”
  • Speaker assignments when it matters who made a statement.

Speaker labels group voices; they do not establish people's identities. Rename or correct them only when you can verify the match. The interview transcription guide explains that review in more detail.

Correcting the source is especially valuable when the mistaken word still forms a plausible sentence. A translator can produce fluent English from the wrong Japanese sentence. Fluency alone is not evidence that the audio was understood correctly.

Translate the reviewed record and check coverage

Use the record's Translate menu to choose an available target language. Complete any language-resource prompt, then inspect the bilingual result. If you translated earlier and subsequently changed the source wording, run translation again for the reviewed version rather than assuming that an earlier translation still matches it.

Check two separate questions:

Does each needed passage have a translation? Look for source-only sections between translated passages, especially near the end. A few translated lines do not prove that the entire document succeeded.

Does the translation preserve the meaning? Compare the names, numbers, conditions, and uncertainty in the source. “We might finish on Friday” must not become a definite Friday commitment. Use the audio and authoritative project material when a line remains unclear.

A recording containing several languages needs extra care. Automatic detection does not guarantee correct handling of every switch, and the available translation pair still matters. For a multilingual set of recordings, plan language settings using the batch transcription guide.

Export the document your reader needs

Open the record's Export menu. Under Export content, choose Bilingual to keep the source and available translation together. Decide whether the document needs timestamps and any verified speaker labels, then choose a format:

  • Format: Plain Text; Useful when: You need reusable text for notes or another editor
  • Format: Markdown; Useful when: You want a text-based writing or documentation workflow
  • Format: PDF Document; Useful when: You want a reading copy to review or distribute
  • Format: CSV; Useful when: You plan to inspect the structured output in a spreadsheet

Bilingual export pairs the source with translations that exist. It does not manufacture missing translations. Translation Only can omit passages that have no translated text, so check coverage before sending a translation-only document. If no translation exists at all, the export cannot serve as a translated document.

Open the saved file after export. Check that it contains the intended languages, that the final passage is present, and that names and characters remain readable. Use a filename that distinguishes the source, target language, and reviewed version.

Keep the recording available and control the copies

A useful review package has three parts: the original recording, the reviewed source text, and the translated or bilingual document. Keep their relationship clear. A transcript that refers to a different edit or an unavailable source is harder to verify later.

Local processing reduces the need to upload the recording to a transcription service. It does not determine where you save or send the exported document. A shared drive, synced folder, or email attachment can distribute the text outside your Mac.

Before sharing, check the actual export rather than relying on the application's display. For an interview, confirm that quotations and speaker names are correct. For a work briefing, confirm commitments against written project records. Mark unresolved passages instead of filling them with a confident guess.

Start with one recording and carry it through this complete review and export process. Once the result is useful, repeat the same checks for the rest of the collection.

The current Mac App Store listing describes file transcription, transcript translation, and export. Product behavior and export distinctions were also checked against the project source on October 24, 2026.

Product reference · 2026-10-24

รายการ Mac App Store ปัจจุบัน

Frequently asked questions

Can I translate a recording without playing it aloud?

Yes. File transcription reads a supported local recording directly. It does not require speaker playback or microphone capture. Live captions are a separate workflow for sound currently playing on the Mac.

Can I export the original and translation together?

Yes. Choose Bilingual in the export-content options. Check for passages without translations before sharing; bilingual output preserves available text but does not guarantee that every passage was translated.

Why is my translated document shorter than the source?

Check for a partial transcription and missing translations. Translation-only export can leave out untranslated passages. Compare the final processed timestamp and several source passages with the saved output.

Does changing a source sentence automatically fix its translation?

Do not assume that a translation produced earlier reflects a later edit. Translate the reviewed record again and inspect the changed passage before exporting.

Can this produce English speech from a foreign-language recording?

This workflow produces text. It does not dub the recording or generate a translated audio track.

Does the app support every recording and language pair?

No. The recording needs a decodable audio track, and recognition and translation resources must support the selected workflow. Check the actual pair on your Mac and try a representative file first.

Related guides