Your words, written,
read, and remembered
Lafal is a private desktop assistant for dictation, writing, on-screen reading, and meeting notes. Use local models by default, with your own cloud provider available when you need it.
Free to start · No account required for local transcription
See what's new →Your data stays in your hands
No analytics. No trackers. No Lafal servers. What you say stays under your control.
Local by default
Local models handle dictation, screen reading, and on-device writing on your machine. Your audio is transcribed locally; nothing is sent to Lafal.
Cloud is opt-in
Bring your own provider and credentials. When you choose cloud writing or summarization, only the text needed for that request goes directly from your device to the service you selected.
No analytics or trackers
No telemetry, usage tracking, fingerprinting, or third-party SDKs phoning home.
Your notes stay yours
History, meeting and imported-media transcripts, recordings, and Markdown exports are stored locally and remain under your control.
Write, dictate, listen, remember
One private desktop assistant for voice dictation, screen reading, meeting notes, and better writing. Keep processing local by default, or connect the provider you already trust.
Speech to text
Hold a hotkey and speak, then release. Local Whisper transcribes your words and types them wherever your cursor is — with push-to-talk, toggle, and multilingual model choices.
Proofread
Grammar, spelling, and punctuation corrected right where you wrote it — no rewrites, no rewording, just a cleaner version of your own words.
Concise
Cuts the filler, keeps the point. Shortens your text in place while preserving exactly what you meant to say.
Rephrase
One idea, any voice. Rewrite what you wrote in the tone the moment calls for.
Summarize
Get the gist in seconds. Opens in a tidy popup you can copy from or close the moment you're done.
Fact check
A six-level verdict on any claim — with a plain-English explanation, a source link when one's available, and a one-click Google fallback whenever the answer is inconclusive.
Text to speech & Lafalify
Drag over any region of your screen and hear it read aloud. Local OCR recognizes the text, highlights each row and word as it plays, and gives you pause, skip, speed, volume, and local voice controls.
Meeting recording & summary
Start a meeting from the tray. Lafal transcribes in the background, captures system audio when supported, and saves a titled note with timestamps, summary, transcript, and recording.
Import audio & video
Have a recording already? Choose an audio or video file from Home and process it as a meeting note or a concise summary. Lafal extracts the audio and transcribes it in bounded, timestamped chunks.
Notes that fit your workflow
Keep notes in Lafal's searchable history or export them to a folder you choose. Meeting exports include linked transcripts and recordings; concise summaries can land in a QuickNote folder.
Local-first models
Choose separate local tiers for text, speech, and voices. They download once and run on your machine — no account or API key required. Connect your own cloud provider when you want to, with model lists refreshed on demand.
History that stays useful
Search past text transformations and browse saved meeting or imported-media notes from one desktop app, with configurable retention and local Markdown exports for your own folders.
Simple, honest pricing
Free to use with local models — no account or subscription required.
Free
Everything you need to write, speak, listen, and keep track of what happened.
- Unlimited local dictation and text tools
- Lafalify screen reading with OCR
- Meeting recordings, transcripts, and summaries
- Import audio/video as a meeting or summary note
- Local speech and voice model choices
- Searchable local history and Markdown notes
- Bring your own cloud provider or API key
Enjoying Lafal? Consider buying me a coffee to support development.
Frequently asked questions
Yes. After the one-time download, Lafal's local speech model can transcribe on your device without an internet connection. Local text models and local screen reading can work offline too.