Shortcut
Start where your hands already are.
Hold a key, tap a toggle, double-tap a modifier, or let auto mode listen for speech.
光标工作流
从语音到光标，无需中途查看转录文本。
DuckType 将听写视为输入，而非需要另行管理的文档。开始录音、说出想法、让所选风格自动处理，然后继续在当前使用的应用中操作。
Speech
Talk in a normal sentence.
DuckType records the utterance from your desktop without needing a browser extension.
Style
Apply a style before insertion.
Use a built-in style, custom style, or one-off restyle to shape the result before it lands.
Insert
The text appears at the cursor.
DuckType can paste normally or type keystrokes for remote desktops and paste-resistant apps.
适用所有应用
只要应用支持键盘输入，就能接受你的声音。
DuckType 在操作系统层面插入文本。自动模式对普通应用程序使用剪贴板粘贴，对远程桌面或禁止粘贴的应用程序使用模拟按键输入，因此无需针对每个应用程序进行单独集成。
Chat and AI
Draft prompts, replies, and long-form instructions without breaking flow.
ClaudeChatGPTGeminiPerplexity
Code and terminals
Dictate commit messages, issue notes, shell comments, and agent instructions.
Claude CodeCursorVS CodeTerminal
Notes and docs
Capture rough thinking into the document or note that should own it.
DuckType NotesObsidianNotionApple Notes
Browsers
Fill text fields, support forms, CMS editors, and web apps from the desktop.
SafariChromeArcFirefox
Remote desktops
Use typed keystrokes when remote sessions block clipboard paste.
CitrixWindows AppWindows 365Remote Desktop
Messages and mail
Speak quick replies, then edit in place before sending.
SlackDiscordMailMessages
启动方式
选择一个能融入肌肉记忆的触发方式。
短暂的想法、长篇草稿和解放双手的场景，各自需要不同的操作方式。DuckType 让你将语音听写映射到最适合当下的手势。
Hold to record
Hold Fn while you speak, release when you are done.
holdFn
Start and stop
Tap once for longer dictation, tap again to finish.
tapCtrlSpace
Double-tap
Double-tap a modifier key when a chord would get in the way.
doubleShiftShift
Auto mode
Voice activity detection splits speech and pastes each utterance.
listenVAD
输入到光标的内容
You said
um can you write a short update saying the windows build is live and smart screen is expected on first run
Styled
The Windows build is live. SmartScreen is expected on first run.
Formatted
Windows build is live. First run may show SmartScreen. Click More info, then Run anyway.
输出
原始转录只是第一稿。
风格与技能可在 DuckType 粘贴前对文本进行改写、翻译或格式化。语言设置和词典让输出更贴近您的实际写作习惯。
Regional spelling
Pick the variant once and stop correcting the same words by hand.
color to colour
analyze to analyse
Your vocabulary
Add names, acronyms, and product terms that generic models miss.
+ TanStack Start
+ Cloudflare Workers
+ Acme onboarding
Romanization
Keep original script and readable Latin text close together.
東京 tokyo
你好 ni hao
한국 hanguk
控制权与所有权
选择转录运行的位置和记录存储的位置。
使用本地模型保护隐私，用自己的服务商密钥实现无限制使用，或选择托管 AI 省去繁琐配置。你的听写记录始终可用。
Offline models
Run local transcription on-device when the audio should stay on your computer.
Bring your own key
Use OpenAI, Groq, Deepgram, ElevenLabs, Mistral, or Cloudflare directly.
Hosted AI when useful
Use DuckType-hosted transcription and skills when speed matters more than setup.
Undo the last paste
If a dictation misfires, remove the inserted text without digging through the field.
History on disk
Search past dictations, copy them again, or feed them into another workflow.
100+ languages
Switch language by shortcut and code-switch mid-sentence when that is how you speak.
听写完成后
转录文本可以成为下一步的内容。
听写往往只是输入的第一步。DuckType 可以对结果运行技能、进行翻译、生成文字转语音音频，并将这些输出与原始文本关联起来。
原始转录文本
Please turn this into a launch update, translate it for the French site, then generate an audio version for the changelog.
Skill
launch update
Translate
fr-FR
Speech
audio saved
Run saved skills
Turn a dictation into a summary, bug report, meeting follow-up, commit message, or any prompt you save.
Skill: release note
Translate the result
Send the same text through a translation skill when the output needs to land in another language.
English to French
Text-to-speech
Generate spoken audio from a transcription or skill result when the final asset needs to be heard.
Voice: product update
Keep every output
Dictations, skill results, translations, and generated speech can stay attached to the same source text.
History: source plus outputs
将语音输入到光标所在之处。
下载 DuckType，在你每天使用的应用中进行语音听写。