OpenLess - Open-source AI voice input method that automatically converts spoken language into structured text

What is OpenLess

OpenLess is built on Rust + Tauri + React + TypeScript It is a fully open-source cross-platform voice input application, supporting macOS & Windows. Press and hold the global shortcut to speak and it will automatically complete recording, ASR transcription, and AI polishing, then insert the result directly at the current cursor position; it targets commercial subscription-based voice input tools such as Typeless、Wispr Flow, Lazy, Superwhisper, etc.

Main features of OpenLess

• Global voice input: In the input box of any application (ChatGPT, Claude, Cursor, Notion, email, chat box, etc.), trigger recording via the global shortcut, and after releasing it will automatically process and insert the text.
• AI Prompt mode: The core differentiating feature. It is not just speech-to-text, but automatically organizes scattered spoken language into structured, constrained, contextual AI Prompts that can be fed directly to ChatGPT / Claude / Cursor.
• Multiple output modes: Supports switching between four modes: "Original text," "Light polishing," "Clear structure (AI Prompt mode)," and "Formal expression."
• Dual-mode recording: Supports two recording methods, "toggle" and "press and hold to talk." At any stage, press Esc to cancel.
• Clipboard fallback: If it cannot insert directly at the cursor position, the result will automatically be copied to the clipboard to ensure the content is not lost.
• Dictionary and hotwords: Supports custom dictionaries of proper nouns, product names, and personal names, injected as ASR hotwords and semantically corrected during the polishing stage.
• In-app automatic updates: Implements automatic client update checking and download/install via the Tauri updater.

How to use OpenLess

• Download the installer for the corresponding platform from GitHub Releases：macOS Users download .dmg and drag it into /Applications；Windows Users download .exe and run the installer.
• First launch on macOS: Grant "Microphone" and "Accessibility" permissions in sequence,quit and reopen the application(Accessibility permission requires a restart to take effect); after granting microphone permission on first launch on Windows, confirm in settings that global shortcut listening is activated.
• App settings: Fill in theVolcengine streaming ASRcredentials (APP ID, Access Token, Resource ID) and thepolishing modelcredentials (Ark / DeepSeek / OpenAI-compatible API Key, Model ID, Endpoint).
• Start recording: Place the cursor in any input box (ChatGPT, Claude, Cursor, Notion, email, chat box, etc.) and press the presetglobal shortcutto start recording; supports two recording modes, "toggle" and "press and hold to talk." Speak naturally into the microphone, and OpenLess completes recording and ASR transcription in real time; after releasing, it automatically enters the AI polishing stage and processes the text according to the mode you selected.
• AI polishing: After polishing is complete, the resultis automatically inserted at the current cursor position; if the target application blocks direct insertion, it automatically falls back to copying to the clipboard to ensure the content is not lost. At any stage (including during recording, during polishing, and before insertion), press Esc to cancel the current task.
• Switch between 4 output modes in the main window: "Original Text", "Light Polish", "Structured (AI Prompt Mode)", "Formal Expression", meeting the written expression needs of different scenarios.
• Add proper nouns: Manually add proper nouns, product names, person names, etc. on the "Dictionary" page. The system will inject them as ASR hotwords and automatically perform semantic correction based on context during the polishing stage.
• Developer build: After cloning the repository, enter the openless-all/app/ directory, run npm ci to install dependencies, then run npm run tauri dev to enter development and debugging mode.

OpenLess official website address

• Official website address：https://openless.top/
• GitHub repository：https://github.com/appergb/openless

OpenLess key information and usage requirements

• Positioning: A macOS & Windows cross-platform open-source voice input tool, benchmarked against closed-source subscription products such as Typeless / Wispr Flow / Superwhisper.
• Core mode: Hold the global shortcut and speak → record → ASR transcription → LLM polishing → automatically insert at cursor position; the core differentiator is AI. Prompt mode, which can directly organize spoken language into a structured Prompt.
• Tech stack: Tauri 2 (Rust backend) + React + TypeScript.
• Current version: v1.2, supports two recording methods: toggle and hold-to-talk, 4 output modes (Original Text / Light Polish / Structured / Formal), Chinese-English bilingual UI, and in-app automatic updates.
• Credential requirements: You need to provide your own Volcengine streaming ASR + Ark / DeepSeek / OpenAI-compatible Chat Completions service.
• License：MIT。

OpenLess core advantages

• Fully open source, local-first: The code is public on GitHub, all data is written locally, and it does not strongly bind to any model or ASR service provider.
• Bring your own cloud credentials: Users configure Volcengine ASR + Ark / DeepSeek and other OpenAI-format-compatible Chat Completions services themselves, with controllable privacy.
• Won't answer for you: The polishing model only organizes text and does not perform Q&A, task execution, or project analysis, avoiding "overstepping its role".
• Chinese-English bilingual UI: Supports switching between Simplified Chinese and English, and automatically adapts to the system language on first launch.

OpenLess comparison with similar competitors

OpenLess use cases

• Generating structured prompts for AI tools: Dictate a vague requirement, and OpenLess automatically organizes it into a detailed prompt with constraints and context, then inserts it directly into the input box for use.
• Drafting emails and requirement documents: Removes spoken filler words, corrects punctuation, and reorganizes paragraphs, turning off-the-cuff speech into written expression.
• Writing code comments and PR descriptions: Converts thoughts directly into standardized, concise text and inserts it at the cursor in an IDE or Git tool.
• Any scenario where you "don't want to type but must output text": Such as filling out forms, writing social media posts, taking meeting minutes, etc. Just hold the shortcut key and speak.