A dictation engine built to fit your workflow exactly. Deeply configurable, powerful out of the box, and works across every app on your system.
Fixed from Maddie to Mehdi
Hey Mehdi, can we sync on the Lang chain integration? The Super base migration is done and the tee RPC endpoints are wired. Prizma was throwing errors so I redeployed on vercell with cloud flare in front. century alerts now route to the right channel. off zero still needs SAML config before anthrop pick can review. Let’s ship by Friday.
Unlimited custom words
Add as many proper nouns, brand names, or industry terms as you need.
Auto-learn from corrections
Edit a transcribed word and it’s added to your dictionary automatically.
Smart filtering
Only learns spelling corrections, ignores unrelated rewrites.
Every detail, your way
A dictation engine built to fit your workflow exactly. Deeply configurable, powerful out of the box, and works across every app on your system.
Custom dictionary
Unlimited custom words: Add as many proper nouns, brand names, or industry terms as you need.
Auto-learn from corrections: Edit a transcribed word and it’s added to your dictionary automatically.
Smart filtering: Only learns spelling corrections, ignores unrelated rewrites.
Snippet library
Voice-triggered shortcuts: Say a keyword and it gets replaced by its full content inline.
Templates and signatures: Signatures, follow-up messages, addresses, boilerplate text.
Fuzzy matching: Close enough matches are recognized and expanded automatically.
Multi-language dictation
90+ languages: Dictate in virtually any language, no add-ons needed.
Switch mid-dictation: Speak in multiple languages within the same session. Switch freely without stopping or reconfiguring.
Real-time auto-detection: Automatically detects your language as you speak. Restrict detection to a custom list or allow all 90+.
Audio preprocessing
Noise filtering: Cleans background noise so the model hears only your voice.
Silence removal: Pause as long as you need. Silences are trimmed so the model won’t hallucinate words that were never said.
Volume normalization: Keeps recognition consistent whether you speak softly or loudly.
Cost and latency optimization: Accelerate and compress audio before sending. Lower latency and lower costs when using your own API keys.
Built-in translations
Any to any: Speak in any language, output in any other. No restrictions.
Fallback target: Set a default language to return to after one-off translations.
Keep original: Disable translation and keep the dictation language as-is.
Formatting templates
Custom instructions: Define what each template does: grammar, tone, or output format.
Per-app defaults: Assign a template per app. Slack gets casual messages, ChatGPT and Claude get prompts, Cursor gets code-ready text.
Switch mid-dictation: Change templates on the fly with keyboard shortcuts while you speak.
Enable and disable: Toggle individual templates on or off without losing their configuration.
App-specific behavior
Auto-detect active app: Settings switch automatically when you change applications.
Per-app presets: Configure dictation language, output language, and cleanup template for each app.
Override on the fly: Change any preset during dictation without stopping. Enable or disable presets without deleting them.
Cloud and local modes
Cloud, local and auto: Cloud for speed, local for privacy, auto to switch on the fly.
Always free locally: Local models run unlimited forever. No account needed, no caps. The product never blocks you.
Managed cloud option: Unlimited transcriptions at $5/mo with a 14-day free trial. No credit card required to start.
Bring your own keys: Use your own API keys with supported providers, or wire in custom endpoints. Unlimited usage, no extra cost.
Independent providers: Set transcription and formatting providers separately. Mix cloud and local freely.
Tailored dictation flow
Tap, hold, or hands-free: Hold to talk or toggle on/off. Custom trigger key or combination.
Widget visibility and position: Always visible, hidden, or on invoke. Drag, pin, or follow cursor.
Audio controls: Dictation sounds, system audio mute, and auto-timeout. All configurable.
Clipboard support: Copy output to clipboard in addition to pasting.
Live controls: Switch templates, language, or settings mid-dictation with shortcuts.
Full data privacy
Everything local: Models, settings, dictionary, snippets, templates, preferences. All stored on your machine.
Zero data retention: We store nothing. No audio, no transcriptions, no preferences. Only auth credentials if you create an account.
No account required: Use the app without signing up. Account only needed for managed cloud proxy.
Local data control: Configure audio retention period or wipe all app data whenever you want.
Analytics off by default: Usage analytics are disabled out of the box. Opt in only if you choose to.
