Make sure the full audio file is stored on the iPhone and the engine is ready, then import it and keep the app open while it transcribes. The wait depends on the phone, recording and temperature. A short test is useful, but it will not predict the time needed for a long interview.
What should I check before starting a long file?
Play a little at the beginning, middle and end to check you have the right recording and that it is complete. Download it if it is in cloud storage. Check that the phone has space for the imported audio and the documents you will make.
Finish the first engine preparation separately. It sets up the bundled speech model for the phone and can take several minutes. Once it is ready, try a short file before committing the phone to a full interview or lecture.
Why does the remaining time change?
The phone does the recognition work. Recording length, device, speech model and temperature affect its pace. Older supported phones use a smaller model, so another phone’s result won’t tell you how long yours will take.
A phone can start quickly and slow as it warms. Other work on the device can affect the conditions too. Treat the remaining time as an estimate and leave time afterwards to check the transcript.
Can I switch apps while it transcribes?
Keep Transcribe Audio to Text open during file transcription. iOS limits work in the background, so switching apps or locking the phone can interrupt progress. If the app reports a stopped or failed job, return to it and use the retry control. Before restarting a long file, check that the original plays through and that the phone has storage available.
The first engine preparation is a separate task. It is safe to switch away, but iOS may pause it until you return. For a long file, choose a time when you can leave the phone available and keep the original recording until you have checked the result.
How do I work through an hour of text?
Start with the opening, ending and a few passages in between. Tap words to replay the parts you need and correct the relevant paragraphs. For later reference, write down important playback times in your notes. If you need to search a long document by word, export DOCX or TXT and use your document app’s Find command.
Chapters can give you titles and jump points on a compatible Apple Intelligence device, with supported languages. It needs at least two minutes of audio and 120 words. Check the titles and listen back before quoting: a chapter heading is a navigation aid, not proof that the underlying words are right.
Questions you may have
How long does an hour of audio take?
There is no fixed time for every phone. Use the app’s estimate and allow for the device to slow as it warms, plus time for your review.
Does offline mean it works in real time?
No. Offline means recognition runs on the phone. When recording in the app, the full transcript is made after you tap Stop.
Can I delete the original file after importing?
Keep it until you have checked the imported audio and transcript and saved the exports you need. It is your reference if something needs checking later.
Do long files need another speech model download?
No. The bundled model handles the recording. A longer file means more audio to process, not another model to download.