Zhipu AI Input Method: Intelligent Voice Input and Editing Tools to Boost Writing Efficiency

Zhipu AI Input Method (AutoTyper) is a desktop-based intelligent input tool developed by Zhipu AI, designed to revolutionize the text input experience in human-computer interaction through large-model technology.Leveraging the GLM-ASR speech recognition model and AutoGLM agent capabilities, this software breaks free from the limitations of traditional input methods confined to mere “typing.” Beyond delivering highly accurate speech-to-text conversion with support for mixed Chinese-English input and multiple dialect recognition, its core strength lies in “AI-assisted writing.”Users can generate structured text directly via voice commands, or refine existing content through polishing, translation, and style rewriting (e.g., converting colloquial speech into formal emails or Lu Xun-style prose). More than a keyboard replacement, it functions as a productivity plugin that understands context and possesses reasoning capabilities. Floating above all desktop applications, it establishes a new paradigm of efficient work—where you speak and it writes—making content creation and communication smoother and smarter.

Function List

• High-Precision Speech-to-TextBased on the GLM-ASR-2512 model, it supports ultra-fast real-time speech recognition, accurately processes mixed Chinese-English input, and supports multiple dialects including Sichuanese, Cantonese, and Northeastern Chinese.
• AI-Powered Polishing and RewritingOffers multiple preset writing styles (such as “for bosses,” “for colleagues,” “Lu Xun-style,” “translation-style,” etc.), instantly transforming simple spoken instructions into appropriate written expressions.
• Multi-scenario Copywriting GenerationBuilt-in rich writing scenario templates, including work reports, leave requests, meeting notices, job applications, Xiaohongshu copywriting, and more. Users simply dictate their core requirements, and the AI automatically generates complete documents.
• Real-time Translation and Cross-Language CommunicationSupports direct voice translation into multiple languages including English, French, German, Japanese, and more, facilitating cross-border communication.
• Global Floating and Cross-Application SupportExists as a floating window, supporting direct invocation within any desktop software capable of text input—including WeChat, Feishu, Word, Notion, browsers, and more.
• Long Text ProcessingSupports extended voice input, ideal for compiling meeting minutes and dictating lengthy articles.
  /n

Using Help

The design philosophy of Zhipu AI Input Method is “invisibility and efficiency.” Once installed, it operates as a system-level utility. Below is a detailed installation and operation guide to help you get started quickly:

1. Download and installation

• downloadingVisit the official website https://autoglm.zhipuai.cn/autotyper/The page will automatically detect your operating system (macOS or Windows). Click the “Download Client” button to obtain the installation package.
• Installation (Windows): Double-click on the downloaded .exe Follow the prompts to complete the installation. Once installed, the software will automatically launch and reside in the system tray.
• Installation (macOS): double-click .dmg Drag the “Zhipu AI Input Method” icon into the Applications folder. Upon first launch, the system will prompt you to grant permissions. Navigate to System Settings > Privacy & Security > Accessibility, then check the box for Zhipu AI Input Method to ensure it can control text input. Additionally, enable its access to the recording device under the Microphone permissions.

2. Account Login

After launching the software, log in using your phone number or by scanning the QR code on the screen with WeChat. New users typically receive a certain amount of free trial credits or time (subject to the latest official promotions).

3. Operation of core functions

• Voice Input (Basic Mode)::
  1. Click the cursor inside any input field (such as a WeChat chat window or a Word document).
  2. Press and hold the keyboard shortcut key (default is usually F1 maybe Option (The key can be customized in settings).
  3. When you start speaking, a ripple animation will appear on the screen to indicate that audio is being captured.
  4. After speaking, release the button, and the text will appear on the screen immediately.
• AI Polishing and Rewriting (Advanced Mode)::
  1. Text Selection MethodSelect a segment of text you've already typed with your mouse. A small “AI Polishing” button will appear next to the software's floating icon. Click it, then choose the desired style (such as “More Formal” or “Translate to English”). The AI-generated text will automatically replace the original or copy to your clipboard.
  2. Voice Command MethodWhile holding the shortcut key, you can directly speak commands such as: “Write a leave request to the boss, citing urgent family matters, requesting two days off.” After releasing the key, the AI won't simply type this phrase but will generate a properly formatted leave request email.
• Switch input style::
  On the floating window interface, tap the settings button next to the microphone icon to preset the output style. For example, setting it to “English” will automatically translate your spoken Chinese into English text displayed on the screen.

4. Personalized Settings

Click the settings icon in the system tray or floating ball to:

• Adjust microphone sensitivity.
• Customize the wake-up shortcut key to avoid conflicts with other software.
• Manage commonly used prompt templates to build your own personalized writing assistant.

First-time users are advised to practice the long-press-to-speak feature in Notepad to familiarize themselves with speaking pace and release timing. Typically, 10 minutes is sufficient to fully adapt to this efficient “hands-free” input method.

application scenario

1. Workplace Communication and Reporting
   When replying to your boss's WeChat messages or drafting weekly reports, simply dictate the gist (e.g., “Inform President Wang that the project is delayed by three days due to server issues”). Select the “For Boss” style, and the AI will automatically generate a tactful, professional, and well-structured report text, avoiding awkwardness from inappropriate wording.
2. International Conferences and Correspondence
   Foreign trade professionals or international students handling English emails can dictate their Chinese content directly and set the output to “English email” style. The software not only delivers accurate translations but also automatically adapts to business email formatting and polite expressions, significantly boosting communication efficiency.
3. Self-media content creation
   For Xiaohongshu or short-video creators, inspiration often strikes too quickly to type. With Zhipu AI Input Method, you can swiftly dictate your ideas and select the “Xiaohongshu Style” option. The AI will automatically add emojis, hashtags, and convert text into conversational speech—generating a draft ready for immediate posting.
4. Real-time Meeting Minutes
   When attending meetings as an observer, open your notepad and activate the extended voice input mode. The software transcribes meeting discussions into text in real time. After the meeting, you only need to organize the generated lengthy text logically to produce comprehensive meeting minutes.

QA

1. Which operating systems does Zhipu AI Input Method support?
   Currently, it primarily supports Windows and macOS systems. The official website provides download clients for both platforms.
2. Is this software completely free?
   The basic download and installation of the software are free. However, using AI large model capabilities (such as speech transcription and text polishing) may consume credits. New users typically receive a certain number of credits upon registration (e.g., 2000 credits). Once depleted, credits may need to be purchased through official channels or obtained by participating in activities. For specific details, please refer to the billing instructions within the software.
3. Can it replace my original Pinyin input method?
   It serves primarily as an auxiliary tool. While it can handle the vast majority of input needs, it works best when used alongside traditional Pinyin/Wubi input methods for entering specific, non-standard short words or extremely brief instant messaging replies. It coexists seamlessly with your existing input methods like Sogou or Microsoft Pinyin without any conflicts.
4. How accurate is the speech recognition?
   Based on the Zhipu GLM-ASR-2512 model, it achieves exceptionally high recognition accuracy in quiet environments (with an extremely low character error rate according to official data). Even under conditions with moderate background noise or faster speech rates, its performance surpasses that of traditional offline speech recognition engines.