Haru Voice Notes

Worried about where your meeting recordings go? Haru Voice Notes turns saved audio into text with local speech models on supported devices, then organizes the transcript with on-device AI.

Save first. Review the original. Keep and play the original audio, and review transcripts and summaries with separate controls. On-device language models produce summary drafts and suggested titles. AI results need your review; check them against the audio and transcript.

Check model readiness. Transcription and summaries depend on device, OS, language and model availability. Summaries and title suggestions require an eligible device with Apple Intelligence enabled and its model ready. Model preparation, including first use or a language change, may require internet access. When only the legacy speech path supports a device and language, transcription is limited to recordings of one minute or less.

UI and processing languages are separate choices. The interface supports Korean, English, French, Spanish and Japanese. Choose Korean, US English, French, Spanish or Japanese for transcription and summaries. Actual availability is checked separately for your device, OS, language support and the readiness of each speech and summary model. Changing the UI language does not prepare a processing model or make speech recognition or AI summaries available.

Sync and sharing are separate choices. Turning on iCloud sync can send saved notes, original audio, transcripts, summaries and attachments to Apple iCloud Drive. Turning it off keeps existing cloud copies; device backups follow system settings. Edit a mail draft and choose transcript or audio attachments, then decide to send in the system mail or sharing screen. Your chosen service may transmit or retain that content.

Apple Watch audio and records transfer to the paired iPhone. Transcription and summaries run locally on that iPhone when its device, language and model conditions are met.

Discovery signals