Per-app contextual profiles


Real-time AI dictation for macOS (Windows soon), free and open source — bring your own Soniox key. Speak, your text appears in any app. Rephrase, translate, shift tone — one click.
All your past dictations.
Profiles per app, a dictionary that learns your jargon, voice snippets, multilingual, AI that rewrites on demand. Every piece is yours to configure.













Dikton types wherever your cursor blinks. Mac, web, terminal, IDE.
Average keyboard speed
Dictation + live correction
Custom profiles and workflows per role — each persona has its dedicated page.
“Option+D in Cursor — my profile translates to English and knows every internal service name. A 300-word brief takes a minute instead of eight.”
“8 client calls a day. Dikton transcribes Zoom audio, speakers separated. I paste into Claude — recap, decisions, action items in 30 seconds.”
“Three profiles: Slack softens, investor Gmail tightens, Claude formats as bullets. I switch apps, Dikton switches profile on its own.”
“Off the call, Option+D in HubSpot: situation, blockers, next step in bullets. The personal dictionary knows every account name.”
“A 1500-word PRD in Notion in 8 minutes instead of 30. My PRD profile formats the sections, I dictate in flow, I don't type anymore.”
“My Figma comments and Linear tickets go out by voice. One profile per app, the right prompt applies before paste.”
The app is free. Paste your Soniox key, pay the provider at cost — not a middleman.
per month, forever
per dictated hour, on your key
free to try, no key needed
No. Processing runs through a secure cloud: audio is processed then dropped, text is never kept. Zero retention. Only anonymous analytics (no content) are stored.
Modern macOS (Apple Silicon M1/M2/M3/M4 and Intel) and Windows 11 x64. The Windows beta includes dictation; meeting transcription remains macOS-only for now.
Default is Alt+D (Option+D on Mac): in any app on macOS or Windows, you speak and text pastes at the cursor. Fully customizable, and you can assign a different shortcut per profile (raw, rewritten, translated…).
Dikton adds AI rewriting with Claude (translate, shift tone, format), per-app contextual profiles, a personal dictionary for your jargon, voice snippets, and multi-speaker meeting transcription (macOS). Native dictation only does raw transcription.
Add your terms (lib names, domain jargon, acronyms, clients) to the personal dictionary. Dikton then picks them up reliably, even rare names. Voice snippets work from the first use.
Yes. 60+ languages supported, mid-sentence switching with no break. Useful for international teams or investor briefs that mix English and French.
Dikton is free and open source. Transcription runs on your own Soniox key: a few cents per hour, trial credit included when you create the account. 20 free minutes let you try before creating your key.
Soniox is the transcription engine Dikton uses. You create an account on console.soniox.com (free, trial credit included), generate an API key and paste it into Dikton. It stays in your Mac's keychain — it never goes through our servers.
Dikton adds meeting transcription in LLM-ready format (paste straight into Claude/ChatGPT) — and it is free and open source, where Wispr Flow costs ~€14/mo.
See the full comparison →Dikton includes contextual AI rewriting on your own Anthropic key plus meeting transcription in LLM-ready format — free and open source, you only pay your actual usage.
See the full comparison →
Free and open source. Bring your own keys — no card, no subscription.