What is on-device recording?
On-device recording is meeting capture that happens on your own computer rather than in a vendor's cloud.Your machine hears the call audio directly, so nothing joins the meeting to record it. Wispr Notetaker works this way, which is why it captures Zoom, Google Meet, Teams, Webex, or a Slack huddle without joining as a bot.
How on-device recording actually works
On-device recording splits listening from transcribing. The listening happens locally, on the machine you are sitting at, which picks up your microphone and the system audio coming out of the call. Two audio streams are involved: your microphone for your side, and the system audio your computer is already playing for everyone else's, combined into one recording.
The alternative is a cloud recorder. There, a participant is added to the call, usually a bot with a name in the attendee list.
Because capture comes from the machine, Wispr Notetaker does not need each meeting platform to let a bot in. It records what the machine plays, so it does not matter which platform the call is on. Nothing joins the call, so the meeting looks normal. Tell everyone on the call that you're recording.
On-device recording vs cloud recording
The split is about where the audio is captured, not where it is turned into text. Most on-device recorders, including Wispr Notetaker, still send audio to the cloud to produce a transcript. Capture happens on your machine rather than on a vendor's servers, so no extra attendee appears in the list. In-person meetings on a shared mic are covered too, and calls that were never on your calendar can be started in one tap. With Wispr Notetaker, if the connection drops mid-meeting, recording continues, and the offline stretch is transcribed after the meeting once you are back online. With a cloud recorder, coverage stops at the platforms the vendor has integrated.
Neither approach is automatically more private. That depends on what happens to the audio next.
On-device recording, on-device transcription, and local-only storage
On-device recording is not the same as on-device transcription. Capture on your machine still usually means the audio is processed in the cloud. Local-only storage is different again. It means the transcript stays on your machine.
The difference matters when a policy asks where data lives. Capture on your machine says nothing about where the transcript is processed, and where it is processed says nothing about how long it is kept. Say you record a call on your laptop with no internet. The audio is captured locally, and the transcript is produced after the meeting once you are back online. Capture was on-device; transcription was not.
What happens to the audio after the meeting ends?
Where the audio goes next matters more than whether a bot showed up in the attendee list. With Wispr Notetaker, meeting audio is encrypted and kept only temporarily, on your device and in some cases in cloud storage, to build your transcript, let you resume a meeting, check transcription quality, and fix problems when they come up. After that limited period it is removed automatically.
Transcripts are a separate setting, and the choice is yours. Keep them indefinitely, which is the default, or have them auto-delete after 1, 7, 30, 90, 180 or 365 days. In an organization, an admin can set that policy centrally, and the setting then shows as managed by the admin.
Wispr Notetaker does not create voiceprints or biometric profiles of you or anyone on your call. Your content is never sold. For more detail, see our Data Controls page.
When on-device recording is the only option a policy allows
Local capture is the only workable option in a few recurring situations.
- Calls on platforms where inviting a recording bot is not an option, since capture comes from your own machine
- Client or candidate conversations where an extra name in the attendee list changes how openly people speak
- In-person meetings and hybrid rooms, where there is no call for a bot to join at all
- Unreliable connections, where audio has to survive the network dropping mid-sentence
A security review usually wants to know two things: whether a bot joins the call, and what happens to the audio afterward. On-device capture covers the first. Retention and model-training settings cover the second.
How Wispr Notetaker handles on-device capture
Wispr Notetaker records on your Mac. One click in the notification joins the meeting and starts recording. Recording length is capped at two hours by default, and you can set 30 minutes, one hour, two, or three.
Losing your connection does not lose the meeting. If transcription cannot connect when a meeting starts, you get the choice to record offline. If the connection drops mid-meeting, recording keeps going and the transcript marks where the drop happened, then live transcription picks up again on its own. Offline audio is transcribed and refined after the meeting, and speaker attribution shows up there rather than in real time.
You see a live transcript during the meeting, with speakers as You and Them. After the meeting, the refined transcript names them. Anyone it cannot identify, you name once and it applies across the whole transcript.
Who on-device recording suits, and who it does not
It suits people who want their own record of their own meetings. A founder in back-to-back external calls. A researcher who needs the customer's exact words. Anyone feeding transcripts into Claude or ChatGPT, where one wrong speaker label spoils the output. It also suits in-person and hybrid work, where a shared mic is the only sensible option.
It fits less well in a few cases. If your team's requirement is that recording is administered centrally, with the organization holding the recording rather than the individual, a cloud recorder matches that shape more directly. Wispr Notetaker runs on Mac today, with Windows coming soon, so a Windows-first team has nothing to roll out yet. Enterprise support is also still to come. And Wispr Notetaker is not available on an account with a signed HIPAA BAA, so a clinical team operating under one is not covered.
Record one real meeting and read the refined transcript against what you remember saying. Record your next meeting with Wispr Notetaker.
Frequently asked questions
Can on-device recording capture an in-person meeting?
Yes. Because capture comes from the machine rather than a call, a shared microphone works. Wispr Notetaker records in-person meetings that way. Speakers show as Speaker 1 and Speaker 2 until the refined transcript resolves them, which happens after the meeting rather than live.
What languages does Wispr Notetaker support?
English today, with more languages coming soon. You can use it on meetings in other languages and expect mixed results until dedicated support arrives. The summary comes back in whichever language was spoken, and a mixed-language meeting follows the language that dominated. Section headings and speaker labels stay in English.
Can I delete a meeting recording after the fact?
Yes. Deleting a meeting is permanent and applies on every device you are signed in on. It takes the notes, summary, transcript, and speaker-identification data with it. Deletion is per meeting, so the rest of your history is untouched.
Do participants know I'm recording if no bot joins?
Nothing joins the call, so the meeting looks normal and recording happens on your device. Tell everyone on the call that you're recording. Wispr Notetaker keeps a "get consent when transcribing others" notice in front of you the whole time it records, your organization can require you to confirm consent before recording starts, and you can add a Wispr Notetaker notice to your calendar invites.
Does Wispr Notetaker train AI models on my meetings?
You choose during onboarding, and you can change it at any time in Settings, under "Improve the model for everyone." On Enterprise it's off for everyone by default. The setting covers dictation and meetings together, so changing it applies to both.
