Can I upload audio files to Wispr Flow for transcription?

Last updated: October 9, 2026

Available on: Mac, Windows, iOS, Android

No. Wispr Flow transcribes live speech from your microphone. You can't upload, attach, or import an existing audio or video file for transcription on any platform or plan, including Voice Memos, meeting recordings, .mp3 and .m4a files, or recordings stored in cloud services.

What Flow captures instead

  • Dictation: Requires microphone access; Mac also requires Accessibility permission. The iOS keyboard additionally requires Full Access. See Set up the Flow keyboard on iPhone.

  • Live meetings: On Mac and Windows, Notetaker captures microphone and system audio during the meeting. Afterwards, Flow re-transcribes its own recording to produce a refined transcript and summary—not a recording you supply.

iOS permission prompts

The keyboard checks the weekly word limit before Full Access, then microphone permission. Without Full Access, it shows “Almost there!” rather than a microphone prompt.

With Full Access granted, denied microphone access shows “We can’t hear you” and “To use Flow, turn on Microphone in Settings”. The “Go to Settings” button opens Flow's own settings, not the iOS microphone controls. Tapping it also lets Flow recheck microphone access after permission is granted.

Warning: On Mac and Windows, speech still awaiting transcription when recording ends can remain untranscribed. Flow can show an error or insert an incomplete transcript. Review the result before relying on it.

Import existing meeting transcripts

On Mac and Windows, imports require a signed-in account with the corresponding import option enabled. The importer is under Settings → Notetaker → Meeting import, using Import or Import again.

  • Otter: Accepts a bulk-export .zip or its extracted folder containing .txt or .srt transcripts—not loose .txt or .srt files.

  • Granola: Accepts a .csv or a .zip containing the export's CSV files.

Otter exports should use TXT with speaker names and timestamps. Flow prefers .txt; .srt is the fallback if .txt is missing, empty or unreadable. Summaries must end in _summary.txt and have a matching transcript. Audio entries in exports are ignored, not transcribed.

Only one import can run at a time. Finish or cancel it before starting another, and wait for an Otter archive to finish reading before selecting another. Each selection can be imported once; reselect the export after 30 minutes, sign-out or restarting Flow.

Is my dictation audio stored by Wispr?

Mac, Windows, and iOS upload history records to Wispr when signed in and online. With sharing enabled, these uploads include dictation audio and text.

The sharing control is Improve the model for everyone on Mac, Windows, iOS and Android. Find it under Settings → Data and Privacy on desktop or Settings → Data and privacy on iOS. Older app versions may label this setting Privacy Mode, with the opposite on/off meaning.

For routine history uploads, audio and transcript text are excluded when Improve the model for everyone is off, including when an organization policy or a HIPAA BAA locks it off. To keep dictation content off Wispr's servers, turn off both Improve the model for everyone and Dictation cloud storage. Metadata can still include word count, duration, app name and URL, timestamp and timezone, microphone device, app version, detected language, edit information, personalization settings, and performance and transcription-quality information.

A signed HIPAA BAA locks sharing and Dictation cloud storage off on Mac, Windows and iOS; organization policies can lock either control. HIPAA or organization policy can also lock Android sharing off. The desktop Dictation cloud storage control appears only when available to your account. Some organization locks can only be changed with Wispr support.

Desktop feedback can include text from the app you're using for the selected transcript only—not during routine history uploads. Desktop history uploads never include screenshots.

Size limits can prevent routine history uploads on desktop, and iOS can skip an oversized recording entirely. A long recording's audio may not be stored even when the rest of its history record is kept. When you report a dictation from History on Mac or Windows, it is sent on its own whatever its length; a very large report sends compressed audio only, and a failed report is retried about 10 minutes later without holding up your other history uploads.

Warning: Submitting transcript feedback on iOS uploads the transcript text and audio regardless of your Improve the model for everyone setting. Feedback submissions are an exception to routine-history restrictions, including organization policies and HIPAA. Wispr retains them in full and can use them to train and evaluate its models.

See Wispr's data usage information.

History and local storage

Your own past dictations stay retrievable from History on Mac, Windows, iOS, and Android.

Warning: Local retention settings and storage limits affect what remains available:

  • Mac and Windows: Flow clears local audio from already-uploaded history records older than 14 days, but keeps transcript text. Records not yet uploaded keep their audio. An organization policy can prevent history and audio from being saved locally.

  • iOS: Auto delete transcripts removes previous days' transcripts only when enabled; it is off by default, but an organization can lock it on. A stricter no-local-storage policy deletes all transcripts except those still in progress. If Flow cannot load your organization settings, it disables sharing and Dictation cloud storage, enables automatic transcript deletion, and locks these settings.

  • Android: History is capped at 100,000 entries with at least 1 GB free, 50,000 with 500 MB to under 1 GB, and 25,000 below 500 MB. Low-storage warnings appear below 1 GB and below 500 MB.

Android History retry uses only Flow's own saved recording and requires its local audio file—not a file you choose to upload. If that audio is missing, Flow shows “Retry failed. Audio file missing.”

FAQs

Will audio file upload be added?

We don't have a timeline to share. Submit a feature request at support.wisprflow.ai, which reaches our support team.