Is Keyspoken an alternative to Wispr Flow or Superwhisper on iPhone?

It does the same core job, turning speech into text in any app, in a different way: with no letter keys, on-device only, using open models and no account. Whether it suits you better depends on whether you want those things or want AI rewriting and a full keyboard.

Is Keyspoken open source?

The speech models it runs are open: NVIDIA Parakeet TDT under CC BY 4.0, through the Apache 2.0 FluidAudio runtime. The Keyspoken app itself is not open source.

Is Keyspoken better than Apple dictation?

We have no measured comparison with Apple dictation, so we make no claim. It is different: an open on-device model, custom vocabulary, whole-dictation Undo and a zero-touch routine. Try both.

Wispr Flow and Superwhisper alternative for iPhone

What Keyspoken does differently

It has no letter keys. It does not replace your keyboard.
It transcribes only on the device. There is no cloud option to turn off, because there is no cloud.
It runs open models: NVIDIA Parakeet TDT, published under CC BY 4.0.
It needs no account and collects nothing.
It finishes by itself: it stops when you pause and returns you to your keyboard.

Keyspoken offering held text with Insert and Discard

Where another app may suit you better

You want AI to rewrite, summarize or change the tone of what you said. Keyspoken types your words as spoken.
You want one keyboard that does both typing and dictation. Keyspoken deliberately leaves typing to Apple's keyboard.
You need Mac, Windows or Android. Keyspoken is for iPhone.
You want to transcribe meetings or audio files. Keyspoken types at the cursor and keeps no recordings.

Five questions to ask of any dictation app

Where is my audio processed: on the phone, or on a server?
Does it replace my keyboard?
Does it work with no connection?
Do I need an account?
What does it cost, and is that a subscription?

Keyspoken and Apple's built-in dictation, side by side

Source for the built-in steps: Apple, iPhone User Guide, Dictate text on iPhone support.apple.com, read 13 November 2026.

	Keyspoken	Apple's built-in dictation
Where speech is processed	On the iPhone, always	On the iPhone in many languages, per Apple
Where it lives	Its own strip. You switch to it with the globe key.	A microphone key on Apple's keyboard. The keyboard stays open.
How it stops	When you pause. Two seconds by default.	Tap the microphone, say "Stop dictation", or 30 seconds of silence
Undo	One tap removes the whole dictation, for two seconds	Say "undo"
Languages	English, or 25 European languages with the Multilingual model	Many. It follows the keyboard language.
Speech model	Open NVIDIA Parakeet models	Apple's own
What you install	The app and a model of about 440 MB	Nothing. It is built in.
Price	Free. No subscription.	Free. It comes with the iPhone.

Compared with the iPhone's built-in dictation

Apple's dictation is built in and needs nothing installed: tap the microphone key on the keyboard and talk. For many people that is the right answer.

Keyspoken is a separate strip that runs a different, open speech model. It adds custom vocabulary for names and terms, one Undo for a whole dictation, and a routine that stops by itself. The fair test is to try both on the things you actually say.

A note on fairness

We do not publish claims about other products. For the speech models themselves, the models page cites the makers' published benchmarks. Our own timings on two iPhones are on the models page. We have not measured accuracy ourselves. For what another app does, check that app's own site.

A different kind of dictation app