Features - AI Voice Input & Text Polishing
Powerful features that make voice input actually useful
Choose features by workflow
The core product is not just speech-to-text. It is a desktop workflow for dictation, AI polishing, Ask Anything, BYOK, and transparent cloud words.
Context-aware writing
Use optional AI polishing with local app classification so the same thought can become an email, chat reply, document paragraph, or technical note.
Ask Anything
Record one spoken question, transcribe it, send the text to your model, and show the answer in a small floating note without chat history.
AI text polishing
Turn rough dictated speech into clearer emails, prompts, docs, notes, and comments before insertion.
BYOK provider control
Bring your own STT and LLM keys when you want direct control over latency, privacy, language support, and provider billing.
Local vs cloud STT
Choose hosted cloud providers for convenience or local/private paths when control and privacy matter more.
Custom dictionary
Improve names, acronyms, product terms, technical jargon, and repeated phrases across dictation and polishing.
Cloud words clarity
Hosted STT and LLM usage can count toward cloud words. BYOK and local provider paths depend on your own provider account.
Developer voice workflows
Use the same hotkey workflow for Cursor, VS Code, Claude, Warp, browser prompts, notes, and code-review comments.
Context-aware writing
Context adaptation is optional and applies only when AI polishing is enabled.
OpenTypeless classifies the active writing surface local, maps it to one of 71 built-in profiles, and falls back to General when unknown. Raw process names, window titles, browser hosts, URL paths, queries, fragments, page content, dictated text, and selected text are not sent as app context.
Tell Sam I can send the Monday update after lunch and include the blocker list.
Hi Sam, I can send the Monday update after lunch and include the blocker list.
Sam, I can send the Monday update after lunch with the blocker list included.
The Monday update will be sent after lunch and will include the current blocker list.
Send Monday update after lunch; include blocker list.
Bring Your Own Providers
Configure your own STT and AI polish providers with your API keys. BYOK STT + 11 LLM providers supported. Switch anytime without losing your settings — no vendor lock-in.
OpenTypeless supports Deepgram, OpenAI Whisper, Groq Whisper, GLM-4-Voice, AssemblyAI, and SiliconFlow as speech-to-text providers. Compare them with recordings in your own languages, accents, vocabulary, and noise conditions, then review each provider's current documentation for streaming support, data handling, regional availability, and billing terms. You can switch providers from the settings panel without rebuilding the rest of your workflow.
AI-Powered Text Polishing
Connect to 11 LLM providers to automatically refine your transcriptions. Fix grammar, improve formatting, adjust tone — all in real-time as you speak.
Supported LLM providers include OpenAI GPT, DeepSeek, Google Gemini, Anthropic Claude, Moonshot, Qwen, Zhipu GLM, Yi, Ollama (local), LM Studio (local), and Groq. The polishing step runs after transcription to fix punctuation, remove filler words, correct grammar, and format your text for professional use. You can customize the polish prompt to match your writing style or add translation instructions.
99 Languages with Auto-Detection
The best STT provider depends on your languages, audio conditions, latency needs, formatting expectations, and data policy.
The best STT provider depends on your languages, audio conditions, latency needs, formatting expectations, and data policy. Record a representative sample, compare the transcripts, and consult the current documentation and billing terms for Deepgram, Groq Whisper, OpenAI Whisper, AssemblyAI, and any other candidate. You can switch providers without losing the rest of your settings.
Global Hotkey — Works Everywhere
Use Typeless-style defaults for fast voice input: Fn on macOS, Right Alt on Windows, and Ctrl+/ on Linux. Ask Anything has its own shortcut.
The global hotkey works system-wide across text editors, browsers, chat apps, email clients, IDEs, and more. Defaults are macOS Fn, Windows Right Alt, Linux Ctrl+/. Ask Anything defaults to Fn+Space, Right Alt+Space, or Ctrl+.. Linux uses Ctrl-based defaults because desktop environments, especially Wayland, vary in how reliably they expose Right Alt.
Custom Dictionary
Add technical terms, product names, and specialized vocabulary. The AI learns your jargon and gets it right every time.
The custom dictionary is especially useful for developers, medical professionals, legal experts, and anyone who works with specialized terminology. Add company names, technical acronyms, product names, or domain-specific jargon. The dictionary entries are sent to the LLM as context during the polishing step, ensuring that terms like 'Kubernetes', 'PostgreSQL', or your company's product names are always spelled correctly.
Privacy First
OpenTypeless can use a configured Custom Whisper-compatible endpoint for local or self-hosted STT and Ollama for the separate local LLM polishing route. Network behavior depends on both endpoints and model dependencies. Check the current STT/LLM documentation rather than assuming its local-processing support or network requirements are fixed.
OpenTypeless can use a configured Custom Whisper-compatible endpoint for local or self-hosted STT and Ollama for the separate local LLM polishing route. Network behavior depends on both endpoints and model dependencies. Since STT/LLM also offers local-processing options, compare its current documentation and verify both products under your intended configuration.
Who uses OpenTypeless?
Start from the work you actually do. Each guide explains the useful workflow and the checks that still belong to you.
Accessibility workflows
Review hotkey remapping, adaptive-switch use, screen-reader handoff, and where manual correction may still be needed.
Students and study notes
Plan a workflow for drafting your own lecture notes, essays, questions, and revisions without promising automatic accuracy.
Business writing
Dictate email and report drafts while keeping provider terms, company data rules, and final human review in the workflow.
Remote work
Use one desktop workflow for chat replies, email, meeting follow-ups, and documents across common collaboration apps.
Journalists
Draft interview notes and story structure by voice, then verify names, quotations, attribution, and source material yourself.
Legal professionals
Explore first-draft workflows with explicit human review of citations, legal language, confidentiality, and provider handling.
Translators
Combine language routes, terminology dictionaries, CAT tools, and endpoint selection for a reviewed translation workflow.
Five precise voice actions
TalkMore routes Draft, Rewrite Selection, Translate, Ask, and Search. It is not a general desktop-control command system.
Draft
Insert a new drafted message at the cursor.
Rewrite Selection
Replace selected text after a supported rewrite request.
Translate
Translate dictated or selected text into the requested language.
Ask
Show a floating-note answer for an informational request.
Search
Open Google, YouTube, Amazon, or GitHub only after explicit search intent.
Frequently Asked Questions
Common questions about OpenTypeless features and capabilities
Do I need an internet connection to use OpenTypeless?
Network use depends on both configured routes. A cloud STT provider sends audio to its endpoint, while a cloud LLM provider sends transcript text for polishing. For local or self-hosted STT, configure a Custom Whisper-compatible endpoint. Ollama and LM Studio are local LLM options for text polishing, not speech recognition. Inspect both endpoint URLs and test the resulting network behavior for your configuration.
Which STT provider should I choose?
The best STT provider depends on your languages, audio conditions, latency needs, formatting expectations, and data policy. Record a representative sample, compare the transcripts, and consult the current documentation and billing terms for Deepgram, Groq Whisper, OpenAI Whisper, AssemblyAI, and any other candidate. You can switch providers without losing the rest of your settings.
Can I use OpenTypeless for translation?
Yes. You can speak in one language and have the AI polish step translate your text into another language. For example, speak in Chinese and get polished English output. Configure this by adding translation instructions to your custom polish prompt in the settings.
Is OpenTypeless really free?
The desktop app is open source under the MIT License. When you bring your own credentials, any STT transcription charges and LLM polishing charges come from the providers you configure. Review each provider's current terms before choosing a route. Managed service options, when offered, are described separately from the MIT desktop app.
What platforms does OpenTypeless support?
OpenTypeless runs on Windows 10 or later (64-bit), macOS 12 Monterey or later, and Linux (Ubuntu 20.04+, Fedora 36+, Arch Linux). The app is built with Tauri, which provides native performance on all three platforms. You can also build from source if your distribution is not officially supported.