Speakli: Voice Typing Keyboard

One big Record button replaces the whole keyboard. Speech becomes clean, punctuated text in any app — offline on your iPhone, with no account and no subscription.

Speakli is a keyboard you talk to.

Where every other keyboard gives you a tiny microphone next to the space bar, Speakli gives you one button the size of the keyboard. Tap it, say what you mean, tap again. Clean, punctuated text lands in whatever app you were already typing in — Messages, Mail, Slack, Notes, a browser field, anywhere.

WHAT MAKES IT DIFFERENT

• One oversized Record button. No hunting for a small target with your thumb while you are thinking about what to say. • You see the words as you speak them. The transcript scrolls like a teleprompter, newest words always in view. • The text arrives finished. Filler words, false starts and missing punctuation are cleaned up before the text is inserted, so you are not editing after dictating.

RUNS ON YOUR IPHONE

Speakli includes two on-device speech engines that work with no internet connection and send nothing anywhere:

• Speakli Engine — Parakeet v3, 25 languages with automatic detection, about 483 MB. • Whisper — Whisper large-v3-turbo through WhisperKit, 99 languages, the best choice when you mix languages inside one sentence. Compressed (646 MB) and full (1.6 GB) builds, your choice.

Apple's built-in recognition is there too, free and instant. If you would rather use a cloud service, you can add your own OpenAI key — Speakli never has a server of its own, and there is nothing to sign up for.

AN AI ASSISTANT THAT WRITES IT THE WAY YOU WANT

Dictated speech is not written language. Speakli can run every dictation through a language model with a prompt you choose:

• Fix spelling and punctuation • Clean up speech — remove filler words, false starts and repetitions • Make it concise • Formal tone or friendly message • Translate to English or Russian • Turn it into bullet points • Write a complete reply to an email

Write your own prompts too, and switch between them straight from the keyboard. Bring your own key for OpenAI, Anthropic, Google Gemini, or any OpenAI-compatible endpoint, including a model running on your own machine.

MORE THAN A KEYBOARD

• Notes — record longer thoughts inside the app. The audio is kept, the transcript is saved, and you can re-process it later with a different prompt. • History — everything you dictated, grouped by day, searchable, with both the raw and the processed text. Copy either one, export the lot as text or JSON. • Dashboard — how much you dictated, how many words, how fast it was processed, over a week, a month or all time.

PRIVACY

The on-device engines never send your audio anywhere. The keyboard needs Full Access only so it can talk to the Speakli app on the same iPhone — that is the only way iOS lets a keyboard use the microphone. Your history and notes stay on your device. There is no account, no analytics, no server run by us. If you turn on the AI assistant or the OpenAI engine, text or audio goes to that provider under your own API key, and Speakli tells you plainly on the screen where it is set up.

FREE, NO SUBSCRIPTION

Speakli costs nothing and has no in-app purchases. If you use a cloud provider, you pay that provider directly for what you use.

Requires iOS 17 or later. On-device engines need an iPhone with a Neural Engine and a few hundred megabytes of free space for the model you choose.