Granola, the AI meeting-notes app, has expanded beyond meetings with a new voice typing feature for Mac. It lets users speak into almost any app and receive cleaned-up, formatted text where the cursor is active.
The feature launched on October 29 and is available to Granola users at no extra charge with unlimited words, according to the company. Granola says versions for Windows, iOS, and Android are coming soon, but it has not announced specific release dates.
This is not a hands-on review. Granola’s speed and accuracy claims still need independent testing across different accents, microphones, languages, and noisy environments. Even so, the launch is useful because it shows how quickly AI meeting assistants and dictation apps are converging into the same category.
How Granola voice typing works
On Mac, users press the Fn key and start speaking. Granola converts the speech into text, adds punctuation and paragraph breaks, then inserts the result into the app currently in use. That could be an email composer, Slack message, note, document, or prompt box for an AI assistant.
Granola’s official voice typing page says the tool is designed for speech that is more natural and less carefully structured than typed prose. Instead of delivering a raw transcript word for word, it attempts to turn the spoken input into polished writing.
That distinction matters. Traditional operating-system dictation is mainly a speech-to-text replacement for the keyboard. Granola is also editing while it transcribes, which can save time but introduces another judgment layer between what the user says and what appears on screen.
It works across apps, not only inside Granola
The most practical part of the release is system-wide access. Users do not have to dictate inside a dedicated Granola document and copy the result elsewhere. The company demonstrates voice typing in email, Slack, Google Docs, notes apps, and AI prompt fields.
TechCrunch reported on October 29 that the Mac feature posts its final text into any app and that mobile and Windows versions are planned. That makes Granola a direct alternative to dedicated AI dictation products such as Wispr Flow, even though Granola built its reputation as a meeting notetaker.
Granola says voice typing is free and unlimited
Granola says voice-to-text is free for all Granola users and does not impose a word limit. That is separate from the company’s broader meeting-note plans, where the free Basic tier limits how much meeting history remains accessible.
There is no stated per-minute charge for the new dictation feature. However, “free” does not answer every long-term pricing question. Granola could change plan limits later, and it has not detailed whether advanced models, team controls, or future platform versions will differ by subscription. Users should check the current pricing page before making the tool central to a workflow.
The company claims typing can be four times faster
Granola markets the feature with a “4x faster” claim based on speaking at 160 words per minute versus typing at 40 words per minute. That comparison describes an idealized input rate, not the total time required to produce accurate final copy.
Real productivity will depend on recognition quality and how much editing is needed afterward. Dictation can be much faster for first drafts, detailed prompts, or routine replies. It can be slower when the subject contains code, unusual names, formulas, precise punctuation, or information that must be reviewed carefully.
Granola says it can draw on vocabulary it already knows from a user’s meetings, including colleague names and team jargon. That may improve recognition for specialized terms, but it also makes privacy settings and account security more important because the product is using workspace context outside the meeting-notes screen.
Privacy is a key buying check
Voice typing captures what a person says and sends it through a transcription workflow. Granola’s security documentation says its products use transcription providers including Deepgram and Assembly and AI providers including OpenAI and Anthropic. The company says audio is deleted after transcription rather than retained as a recording.
That means users should not assume the new feature is fully on-device simply because it runs from a Mac app. People handling confidential work should review Granola’s privacy policy, subprocessor list, retention controls, model-training settings, and employer rules before dictating sensitive material.
The risk profile is also different from meeting capture. Dictating a personal email draft usually records only the user’s own speech, while running Granola during a meeting can capture other participants. Granola offers notification and consent features for meetings, and local laws or workplace policies may require participants to be informed.
What voice typing can and cannot replace
For everyday writing, Granola could reduce the friction of getting a rough idea onto the screen. It is especially relevant for long prompts, status updates, meeting follow-ups, and first-pass drafts where speed matters more than exact formatting.
It is less suitable as a blind final-output system. Names, dates, prices, links, technical terms, and commitments should still be checked before sending. AI cleanup can also change tone or remove hesitation that carried meaning in the original speech.
Granola has not published an independent accuracy score for the launch, a complete list of supported voice-typing languages, or dates for Windows and mobile availability. The company’s meeting product supports multiple languages, but buyers should not assume identical coverage or quality for this new feature until Granola confirms it.
Why meeting apps and dictation tools are converging
Granola’s move follows the reverse path taken by some dictation competitors. Wispr, known for voice input, added meeting-notes features; Granola is now adding voice input to a meeting-notes product. Other speech-focused apps are expanding in similar ways.
The logic is straightforward: the underlying tools—audio capture, speech recognition, vocabulary context, formatting, and AI rewriting—overlap. A company that already handles meeting speech can extend that system to writing between meetings. For users, the benefit could be one account and one vocabulary model across more tasks. The downside is giving a single service access to a wider slice of daily work.
Who should try it
Existing Granola users on Mac have the clearest reason to test voice typing because it is included and uses the context already stored in their account. People who write long messages or AI prompts may also find the system-wide workflow more convenient than recording audio and copying a transcript.
Windows, iPhone, and Android users will need to wait. Mac users should compare the output with Apple’s built-in dictation and other AI voice tools using their own vocabulary and privacy requirements. The best option will be the one that produces accurate text with the least correction—not simply the one with the biggest speed claim.