Max Lin

按住 fn 说话,
松开就是文字
Hold fn and speak. Let go, and the text is there.

文字直接粘贴到光标处,任何 App 的输入框都行。识别用你自己的千问或豆包 Key,没有 Kay 服务器,也没有订阅。The text is pasted at your cursor, in any app. Recognition runs on your own Qwen or Doubao key: no Kay server, no subscription.

下载 DMGDownload the DMG 免费开源,经 Apple 公证。需要 Apple 芯片的 macOS 26,和一个阿里云或火山引擎的 API Key。Free, open source and notarized by Apple. Needs macOS 26 on Apple silicon and an Alibaba Cloud or Volcengine API key.

你的声音只去你选的那一家Your voice goes to one place: the service you chose

Kay 没有自己的服务器。按住键时,音频从你的 Mac 直接发到阿里云千问或火山引擎豆包,用的是你自己的账号和 Key,两家随时切换。Kay has no server of its own. While you hold the key, audio goes straight from your Mac to Alibaba Cloud’s Qwen or Volcengine’s Doubao, on your own account and key. Switch between them any time.

  • 只在按住时录音Recording only while you hold松开键,录音就停,识别结果粘贴到光标处。Let go and recording stops; the result is pasted at your cursor.
  • 纠错只发文字Correction sees text, never audio填了 DeepSeek Key 才会纠错,只改听错的专名和标点,不换说法。3 秒没结果就贴原文。Only with a DeepSeek key added. It fixes misheard names and punctuation without rewording; no answer in 3 seconds and Kay pastes what it heard.
  • 没有中间这一层Nothing in between识别费用由服务商按标准价格直接向你收。开发者收不到你的音频和文字。The service bills you directly at its standard rates. The developer never receives your audio or text.
组合键照常用Shortcuts still work
按住时按了别的键就取消录音;按住不到 0.3 秒算误触。Press another key while holding and the recording is cancelled; a hold under 0.3 s is ignored.
剪贴板原样放回Clipboard put back
粘贴完恢复你原来复制的内容,也可以不进剪贴板历史。What you had copied is restored, and dictations can stay out of clipboard history apps.
历史在本机History on your Mac
最近 1000 次听写可以搜索,最近 20 次留着音频。The last 1000 dictations are searchable; the last 20 keep their audio.
崩溃也不丢Survives a crash
听写中途退出或崩溃,下次启动重新识别那段音频。Quit or crash mid-dictation, and that audio is recognized at the next launch.
平台Platform
macOS
要求Requires
macOS 26 或更高,仅支持 Apple 芯片。需要阿里云百炼(千问)或火山引擎(豆包)的 API Key;纠错可选,需要 DeepSeek API Key。需要授予「麦克风」和「辅助功能」权限。界面有简体中文和英文。macOS 26 or later, Apple silicon only. An API key for Alibaba Cloud Model Studio (Qwen) or Volcengine (Doubao); a DeepSeek API key for the optional correction. Microphone and Accessibility permission. Interface in English and Simplified Chinese.
许可证License
MIT

常见问题Questions

怎样在 Mac 上任何地方用语音输入文字?How do I dictate text anywhere on a Mac?

装好 Kay,允许「麦克风」和「辅助功能」,在「设置 → 语音服务」里选千问或豆包并填入 API Key。之后把光标放进任意 App 的输入框,按住 fn 说话,松开后文字就粘贴在光标处。如果松开 fn 后弹出表情面板,把「系统设置 → 键盘 → 按下 🌐 键时」设为「不执行任何操作」。Install Kay, allow Microphone and Accessibility, then pick Qwen or Doubao in Settings → Speech Service and add its API key. Put your cursor in a text field in any app, hold fn, speak, and let go: the text is pasted at the cursor. If the emoji picker appears when you let go of fn, set System Settings → Keyboard → “Press 🌐 key to” to “Do Nothing”.

Kay 和 macOS 自带的听写有什么不同?How is Kay different from the Dictation built into macOS?

macOS 听写是系统自带的,不需要 Key。Kay 的做法不同:按住键说多久就录多久,松开才粘贴;识别由你选的云服务(阿里云千问或火山引擎豆包)在你自己的账号下完成;可选的 DeepSeek 纠错会在粘贴前改正听错的专名;每次听写都保存在本机可搜索的历史里。macOS Dictation is built in and needs no key. Kay works differently: it records for exactly as long as you hold the key and pastes when you let go; recognition is done by the cloud service you choose (Alibaba Cloud Qwen or Volcengine Doubao) on your own account; an optional DeepSeek pass fixes misheard names before pasting; and every dictation is kept in a searchable history on your Mac.

Kay 用的是哪个语音识别?要花多少钱?Which speech recognition does Kay use, and what does it cost?

二选一:阿里云百炼的千问语音识别(qwen-audio-3.1-asr-flash-streaming),或火山引擎的豆包流式语音识别 2.0。Kay 本身免费开源,没有订阅;识别费用由你选的服务商按它的标准价格直接向你收取:阿里云按输入、输出 Token 计费,豆包按你购买的资源类型(小时版或并发版)计费。DeepSeek 纠错可选,同样由 DeepSeek 收费。Your choice of Alibaba Cloud Model Studio's Qwen ASR (qwen-audio-3.1-asr-flash-streaming) or Volcengine's Doubao Streaming ASR 2.0. Kay itself is free and open source, with no subscription; the service you pick bills you directly at its standard rates: Alibaba Cloud by input and output tokens, Volcengine by the resource you bought (hourly or concurrent). The DeepSeek correction is optional and billed by DeepSeek.

我的录音会发到哪里?Where does my audio go?

只在你按住快捷键时录音,音频从你的 Mac 直接发到你选的服务:你配置的阿里云工作空间地址,或火山引擎的 openspeech.bytedance.com。填了 DeepSeek Key 时,识别出的文字(不是音频)会发到 api.deepseek.com。开发者没有服务器,收不到你的音频和文字。历史和最近 20 次的音频只保存在本机。详见隐私政策。Kay records only while you hold the key, and the audio goes straight from your Mac to the service you picked: the Alibaba Cloud workspace address you configured, or Volcengine's openspeech.bytedance.com. With a DeepSeek key added, the recognized text (not the audio) goes to api.deepseek.com. The developer runs no server and never receives your audio or text. History and the last 20 recordings stay on your Mac. See the privacy policy.

为什么需要「麦克风」和「辅助功能」权限?Why does it need Microphone and Accessibility?

「麦克风」用来在你按住快捷键时录音。「辅助功能」用来在任何 App 里发现你按下了快捷键,并替你执行粘贴。没有「辅助功能」时,Kay 只把文字复制到剪贴板,由你自己粘贴。两项都在 系统设置 → 隐私与安全性 里授予。Microphone lets it record while you hold the key. Accessibility lets it notice the key in any app and paste the text for you; without it, Kay copies the text to the clipboard for you to paste. Grant both in System Settings → Privacy & Security.

打开时会被 Gatekeeper 拦住吗?Will Gatekeeper block it?

不会。发布的 App 用 Developer ID 签名并经过 Apple 公证,下载 DMG 后拖进「应用程序」即可直接打开。只支持 Apple 芯片的 Mac。No. The released app is signed with a Developer ID certificate and notarized by Apple, so it opens normally after you drag it from the DMG to Applications. It runs on Apple silicon Macs only.

怎么卸载?How do I uninstall it?

如果开了「登录时打开」先关掉,再从菜单栏退出 Kay,把它从「应用程序」拖到废纸篓。如需清干净:先在设置里对每个服务点「移除 Key…」(从钥匙串删除;Key 在服务商那边仍然有效),删除 ~/Library/Application Support/Kay(历史和录音)、~/Library/Logs/Kay、~/Library/Preferences/com.max1874.kay.plist 和 ~/Library/Caches/com.max1874.kay,并在 系统设置 → 隐私与安全性 里移除它的「麦克风」和「辅助功能」权限。Turn off Open at Login if you enabled it, quit Kay from the menu bar and drag it from Applications to the Trash. To remove everything: first use Remove Key… for each service in Settings (this deletes it from the keychain; the key stays valid with the provider), delete ~/Library/Application Support/Kay (history and audio), ~/Library/Logs/Kay, ~/Library/Preferences/com.max1874.kay.plist and ~/Library/Caches/com.max1874.kay, and remove its Microphone and Accessibility grants in System Settings → Privacy & Security.

获取帮助Get help

发邮件到Email hi@max1874.com,或在 , or open an issue on GitHub 提 issue。写上设备型号和系统版本,能更快定位问题。. Include your device model and OS version so the problem is easier to find.