project
Metasequoia Input Method - A free and open-source Chinese input method that supports AI predictive text.
Metasequoia Input Method is a free and open-source Chinese input method based on the native Windows TSF framework, designed specifically for Win10/Win11. Unlike most RIME-based solutions, this tool is entirely self-developed and natively compatible with UWP and high-DPI...
Metasequoia Input MethodWhat is it?
Shuishan Input Method is a free and open-source Chinese input method based on the native Windows TSF framework, designed specifically for Win10/Win11. Unlike most RIME-based solutions, this tool is entirely self-developed and natively compatible with UWP and high-DPI multi-screen environments. Shuishan Input Method supports Pinyin, Shuangpin, and Wubi, and can enable voice input and AI predictive text via its own API. Its core input logic runs locally, with no ads or privacy breaches, making it a new choice for Windows users seeking a clean and secure input experience.
Metasequoia Input MethodMain functions
-
Multiple input schemesIt supports three mainstream input schemes: Pinyin, Shuangpin, and 86/98 Wubi.
-
Auxiliary code filteringSupports auxiliary code schemes such as Blue Sky Raindrop and Natural Code, accurately locating duplicate codes.
-
Super Simplified Pinyin and Quick PhrasesSupports custom shortcut phrases to improve the efficiency of inputting frequently used terms.
-
Voice inputTo initiate recording via the right-click menu, you need to configure the ASR API to achieve speech-to-text conversion yourself.
-
AI-assisted associationIt asynchronously generates a third intelligent suggestion when inputting in full Pinyin/double Pinyin, and supports custom suggestion words and LLM APIs such as DeepSeek.
-
Text polishingSupports real-time polishing of input content by accessing the LLM API.
-
Handwriting recognition tabletProvides a mouse/touch handwriting input panel that automatically recognizes candidate Chinese characters.
-
On-screen keyboardSupports text input via mouse or touch and shortcut keys.
-
Emoji and Emoticon PanelBuilt-in emoji input for quick input.
-
Customized dictionarySupports custom import and addition of Pinyin dictionary, Wubi dictionary, and English dictionary.
-
Appearance and shortcut keysSupports dark/light themes, customizable candidate window styles, and shortcut key configurations for page turning/editing.
Follow us on WeChat and reply with "open source",join inAI open source project discussion group
How to use the Metasequoia input method
-
Prepare the environmentInstall Visual Studio 2026, CMake, vcpkg, Python 3.10+, and gsudo (installed via Scoop).
-
Cloning repository:implement
git clone --recursive https://github.com/metasequoiaime/MetasequoiaImeTsf.git。 -
Initialize environmentRun after entering the directory
python .\scripts\prepare_env.pyAutomatically configure dependencies. -
Compiling a DLL: Execute separately
.\scripts\lcompile.ps1 64and.\scripts\lcompile.ps1 32Generate dual-architecture components. -
Deployment filesCreated as an administrator
C:\Program Files\MetasequoiaImeTsfandC:\Program Files (x86)\MetasequoiaImeTsfCopy the corresponding bit version of the DLL from the directory. -
Register input method: Execute in administrator PowerShell
sudo regsvr32 .\MetasequoiaImeTsf.dllComplete system registration. -
Configure APITo enable advanced features, simply enter the interface addresses and tokens for the ASR API (voice) and LLM API (AI association/polishing) in the settings panel.
The core advantages of the Metasequoia Input Method
-
Native TSF architectureIt is deeply adapted to modern Windows applications, with better compatibility than traditional IMM32 and cross-platform solutions.
-
Fully open source and self-developedIt does not rely on the RIME engine, and the code is independently controllable, filling the gap in the pure self-developed TSF open source input method for Windows.
-
Ultimate privacy and securityThe core input logic runs locally, with no cloud-based dictionary uploads, and no pop-up ads or privacy theft.
-
AI capabilities are pluggableBoth voice and AI connectivity are accessed through open APIs, allowing users to choose service providers and models independently, offering flexibility and low cost.
-
Dual architecture compatibilityIt provides both 64-bit and 32-bit components to ensure stable operation in various new and old software.
The address of the Metasequoia Input Method project
- GitHub repository：https://github.com/metasequoiaime/MetasequoiaImeTsf
Application scenarios of Metasequoia Input Method
- Privacy-first office scenariosSuitable for government, finance, R&D and other positions where input data security is sensitive, the core input runs entirely locally, eliminating cloud uploads and privacy theft by commercial input methods.
- Windows native environment deep usersFor Win10/Win11 users who frequently use UWP applications, work on high-resolution screens, or work with multiple monitors, the TSF architecture provides a more stable compatibility and rendering experience than traditional input methods.
- Advanced input methods for Shuangpin and WubiIt supports auxiliary codes for accurate filtering of duplicate codes and super simplified spelling, meeting the needs of professional writers who pursue efficient shape code input and require custom shortcut phrases.
- AI Input Method Experimentation and CustomizationBy providing its own DeepSeek and other LLM APIs and ASR services, it offers a low-cost experience of AI-powered predictive text and voice input, making it suitable for tech enthusiasts who enjoy tinkering with tools.