Can Dragon transcribe audio files?

Yes. Dragon can transcribe a recording you already have, not only the speech you dictate live. On Windows, Dragon 15 has a menu for it. From the DragonBar, choose Tools, then Transcribe Audio, then Transcribe Recording ¹. The fit is good for your own voice ², and workable for one other person if you accept hand corrections. It is a weak fit for interviews with several speakers.

Dragon accepts common audio files, and the Windows and Mac lists differ

On Windows, Dragon 15 lists .wav, .mp3, .wma, .dss, .ds2 and .m4a as supported formats, with a minimum sampling rate of 16 kHz ³. The Mac help lists a different set: mp3, mp4, wav, aif, aiff, m4v and m4a ⁴. If your file is in neither list, convert it before you start.

Check the list against the version you own. The Windows page is the Dragon 15 help and dates from 2018, so a newer release may change the menus or the formats.

The Mac list needs a caution too. Nuance said it would stop providing updates for Dragon Professional Individual for Mac after 22 October 2018 ⁵. Treat the Mac format list as a description of that older product, and confirm it with the current Mac release before you rely on it.

Someone else's voice is where Dragon falls short

When the voice belongs to someone else, Nuance's help says that you can correct the output only by hand, with the mouse and keyboard ⁶. In the Transcribe Recording step, Dragon asks who the voice belongs to, and the two choices are Me or Someone else ⁷. The draft is still a machine draft, so check it line by line.

Dragon also works in a two-step pattern. The University of Sheffield library describes using Dragon to dictate into a digital recorder and to transcribe the recording later ⁸. That helps when you want to speak notes away from a computer. It does not change the fact that the draft depends on whose voice is on the recording.

Dragon is a dictation product, so check the fit before you buy

Dragon is sold as dictation software first, and file transcription is one of its functions. A one-user Dragon Professional v16 license is listed at $685.99 by one US reseller, checked on 16 October 2026 ⁹. That is a one-time price for the software, not a transcription service. Compare it with pay-per-hour services only on the total you will pay for the hours you actually transcribe.

For several speakers, the labels, timestamps and export format decide the result. See speaker diarization for what labels are, and the transcription software guide for how to compare tools on those features.

What to do

1. Open the Dragon version you own and look for Tools, then Transcribe Audio. If the menu is not there, your version differs from the Dragon 15 help, so use the version's own help.
2. Convert the file to one of the formats your version lists. For the Windows Dragon 15 list, keep the sampling rate at 16 kHz or higher ³.
3. Transcribe ten minutes of the hardest audio, then score the errors yourself. The word error rate tells you how many words were wrong.
4. If the recording has several speakers, or the transcript must be exact, test a transcription service on speaker labels and export formats before you decide.

Dragon is not the only option for files. Whether Microsoft Word can transcribe audio covers Word's Transcribe feature, which turns an uploaded file into a transcript with speakers separated. Whether Google Docs can transcribe audio covers the Docs route. For manual typing with a foot pedal, read what Express Scribe is.

For live dictation rather than file transcription, Yap, which we make, runs on the device on Mac and Windows. See offline dictation on Mac. For the full set of tools and how to test them, return to the transcription overview.