先看到可用文字，输入不再是黑盒等待。 See usable text early instead of waiting for a black-box transcript.
iOS 语音键盘 · macOS 快捷输入
iOS voice keyboard · macOS quick dictation
Stream-first Voice Keyboard Stream-first Voice Keyboard
先实时流出，边说边修复，停下即定稿，越用越懂你。 Stream first. Repair live. Finalize clean. Learn your vocabulary.
Echo 把听写变成可控的输入回路：文字先出现，上下文在说话时修复，停下后快速收敛成最终稿，个人词库持续记住人名、项目名和专业术语。 Echo turns dictation into a controllable input loop: live text appears first, context repairs happen while you speak, final text settles when you pause, and your personal vocabulary improves future sessions.
Stream First：先出字Stream First: text appears early
Repair Live：边说边修Repair Live: context fixes in motion
Finalize Clean：停下定稿Finalize Clean: settled after pause
Learn Vocabulary：越用越懂Learn Vocabulary: remembers your terms
Echo 1.1.3
新版本：个人词库识别引擎
New: Personal Vocabulary Engine
你手动保存的名字、项目名和专业术语优先进入识别提示，让 Echo 更像懂你的输入伙伴。
Names, projects, and specialized terms you save now get priority as recognition hints, making Echo feel more personal.
边说边出现Appears while you speak
实时可编辑Live & editable
草稿形成时修复Repair while the draft forms
上下文修复Repair in context
草稿就地收敛The draft settles in place
停下定稿Pause to finalize
示意动画 · 非实机录屏Illustrative animation · not a screen recording
可编辑Editable
边说边审阅文字
Review text while speaking
上下文In context
在形成过程中修复
Repair as the draft forms
停下后After a pause
从已读草稿快速定稿
Finalize from the draft you saw
个人词典Personal vocabulary
记住你的名字和术语
Learns names and terms you save
Echo 主线 The Echo loop
先实时流出，边说边修复，停下即定稿，越用越懂你。 Stream first. Repair live. Finalize clean. Learn your vocabulary.
这不是四个独立功能，而是一条输入路径：先降低等待，再降低重读成本，最后让个人词库把下一次识别变得更准。 These are not four separate features. They are one input path: reduce waiting first, reduce rereading next, then let personal vocabulary make the next session more accurate.
同音词、漏词、边界和标点在说话时就开始收敛。 Homophones, omissions, boundaries, and punctuation start converging while you speak.
停下后快速定稿，不是重新等待整段批处理。 After you pause, Echo settles the text fast instead of starting a batch pass.
名字、项目、术语持续进入个人词库，下一次更贴近你的语境。 Names, projects, and terms feed your dictionary, so future sessions match your context.
任何输入场景 Works everywhere you write
不是另开一个编辑器，而是增强你正在输入的地方。 Not another editor. A faster input layer for the apps you already use.
Messages
Mail
Notes
Docs
Slack
ChatGPT
Cursor
Gmail
Notion
WeChat
WhatsApp
Echo 的差异化 Why Echo is different
不只是语音转文字，而是一条从实时流、实时修复、快速定稿到个人词库的输入主线。 Not just voice-to-text: a complete loop for streaming, live repair, clean final text, and personal vocabulary.
传统听写是“说完、等待、重读整段、再纠错”。Echo 把输入拆成四步：先实时可见，在过程中修复，停下后定稿，再用个人词库让名字、项目和专业术语越说越准。 Traditional dictation is speak, wait, reread, then fix. Echo splits the loop into four parts: stream early, repair in motion, finalize after pause, then use personal vocabulary to improve names, projects, and technical terms.
边说边审阅 Review while speaking
Stream 让你在说话时就理解转写结果，错误更早暴露，不必等整段生成后才开始确认。 The stream gives you useful text immediately, so understanding and correction happen before the paragraph is over.
边转写边修复 Corrects in the stream
上下文会在转写过程中修复许多同音字、漏词和标点；最终稿不是从零开始等，而是在实时结果上收敛。 Context repairs many homophones, omissions, and punctuation issues as the transcript evolves; final text builds on what you already reviewed.
越说越懂你 Learns your language
人名、专业名词、同义词和最近上下文会进入个人词典；你说得越多，修错越贴近你的语境。 Names, terms, synonyms, and recent context feed a personal dictionary, so correction gets closer to your world over time.
实时对比 Live comparison
不是录完再等。Echo 在你说话时就开始修复。 Not record, wait, reread. Echo repairs while you speak.
传统听写把理解和纠错都放到最后；Echo 把文字、上下文修复和最终确认提前到转写过程中，让停下后的结果几乎已经被你审阅过。 Traditional dictation pushes understanding and correction to the end. Echo moves live text, context repair, and review into the speaking moment, so the final result is already familiar when you pause.
示意动画 · 不是同条件速度测试Illustrative animation · timing is not a measured comparison
四步输入主线 Four-part input loop
先实时流出，边说边修复，停下即定稿，越用越懂你。 Stream first. Repair live. Finalize clean. Learn your vocabulary.
Echo 优化的是完整输入回路：用户不需要说完后等待、再花时间读懂和纠错，而是在说话时就开始确认结果。 Echo optimizes the whole input loop: less waiting after speech, less rereading after output, and more confidence while the text is still forming.
L1
Live Stream
开口即出字，先给你可用的理解面。输入不是黑盒等待，而是实时可检查。 Useful text appears as you speak, turning dictation from a black-box wait into a live review surface.
L2
In-stream Repair
上下文会在转写过程中修复同音字、漏词、边界和标点，让草稿不断接近可用文本。 Context repairs homophones, omissions, boundaries, and punctuation while the draft is still forming.
L3
Final + Polish
停下后快速定稿。Quick 只做转写，不经 LLM 改写；需要润色时可选择 Polish。 After you pause, Echo settles the final text fast. Quick is transcription only, without LLM rewriting; choose Polish for AI editing.
动效演示 In motion
先看到实时文字，再拿到已经审阅过的最终稿。 Live text first. Better final text after you pause.
01
实时流 + 快速定稿 Live stream + fast final
用户在说话时已经看过主要内容；停下后只是把结果快速收敛，而不是重新等待整段。 You already reviewed the main text while speaking; after the pause, Echo converges to the final answer instead of starting a new wait.
02
上下文修错 + 智能润色 Context repair + Smart Polish
先根据语境修正常见转写错误，再按需把口语变成更干净的可发送文本。 Repair common transcription errors with context first, then turn rough speech into cleaner ready-to-send text when needed.
核心功能 Core features
用户马上能感受到的四个承诺。 Four promises users feel immediately.
核心是实时可见和边转写边修复；个人词典让它越用越聪明，AI Polish 则在需要时完成更强的二次整理。 The core is live visibility and repair while streaming; the personal dictionary makes it smarter over time, and AI polish handles heavier second-pass output when needed.
Compatible
从兼容的输入框开始。 Dictate in compatible text fields.
iOS 键盘和 macOS 快捷键，在 Messages、Mail、Notes、Slack 等兼容输入框内加入语音输入。 Echo's iOS keyboard and macOS hotkey add voice input in compatible text fields, including Messages, Mail, Notes, and Slack.
Live + Final
边说边看，停下即定稿。 Review live. Finalize fast.
转写时就修复一部分问题，停下后再用整段上下文快速收敛最终答案。 Repair begins during the stream, then full context quickly settles the final answer after you pause.
AI Cleanup
需要时再深度整理。 Polish only when needed.
Quick 专注转写。需要邮件、要点或译文时，再选择 Polish、格式化或翻译。 Quick keeps to transcription. Choose Polish, formatting, or translation when you need an email, bullet points, or another language.
Personal
越说越贴近你的语境。 Learns your context.
你明确保存的人名、项目名和专业术语拥有更高优先级；历史与上下文优先留在本地。 Names, project terms, and domain vocabulary you explicitly save get stronger priority; history and context stay local-first.
产品细节 Product details
四个承诺，展示 Echo 为什么像一层新的输入系统。 Four product promises, shown as one input system.
Product evolution Product evolution
每次迭代，都在缩短“想法到文字”的距离。 Every release shortens the path from thought to text.
Echo 的路线不是堆功能，而是把输入闭环越做越可信：先快起来，再变干净，最后变得更个人化。 Echo is not piling on features. It is making the input loop more trustworthy: faster first, cleaner next, then more personal.
1.1.3
Personal Vocabulary Engine
Personal Vocabulary Engine
Echo 现在更懂你的名字、项目和专业词。 Echo now understands your names, projects, and domain terms better.
手动词典优先、热词提示更稳、学习词清理更安全。你明确保存的词会优先被 Echo 尊重。 Manual dictionary priority, safer hotword hints, and cleaner learned-term maintenance. Words you explicitly save come first.
V4
Fast Stream + Polish Modes
Fast Stream + Polish Modes
速度和表达分层：先输入，再按需变好。 Speed and expression are separated: input first, refine when needed.
Stream 让文字边说边出现。Quick 不运行 LLM 润色；Polish、格式化和翻译按需启用。 Stream makes speech visible as you speak. Quick skips LLM polish; choose Polish, formatting, or translation when needed.
1.0.9–1.1.0
Keyboard Trust
Keyboard Trust
像原生键盘一样可靠，是所有 AI 能力的地基。 Native-feeling keyboard trust is the foundation for every AI feature.
持续打磨拼音、候选词、布局、暗色模式和录音启动，让 Echo 先成为可以每天用的输入工具。 Pinyin, candidates, layout, dark mode, and recording reliability keep improving so Echo can be a daily input tool first.
查看完整版本记录，包括 1.1.3 的个人词库升级。
Read the full release notes, including the 1.1.3 vocabulary upgrade.
打开更新日志
Open changelog
工作流对比 Compare workflows
同一条工作消息，哪种输入方式更适合你？ One work message. Which workflow fits you?
以下是各产品官方说明的摘要，核对日期为 2026 年 9 月 20 日，不是同条件实测或准确率排名。功能随平台、模式和版本变化；先在你自己的 iPhone 上试一条中英混合消息。 A summary of official product descriptions, checked September 20, 2026, not a controlled test or accuracy ranking. Features vary by platform, mode, and version. Try a Chinese-English message on your own iPhone.
| 选择时看什么What to compare | Echo | Wispr Flow | Superwhisper | Typeless |
|---|---|---|---|---|
| 转写与改写Transcription and editing | 先转写，按需润色Transcribe first. Polish optionally. Quick 只做语音转写，不运行 LLM 改写；Polish 是可选的 AI 整理。识别仍可能出错。 Quick transcribes without LLM rewriting. Polish is optional AI editing. Recognition can still make mistakes. | 官方说明包括语气词清理和口述自我修正。桌面版在停止后插入完成文本；不要把桌面行为直接等同于 iPhone。 Documents filler cleanup and spoken self-corrections. Desktop inserts completed text after stopping; that does not establish iPhone behavior. | 模式组合语音模型与可选 AI 处理。Voice to Text 模式不做 AI 处理；其他模式可进一步整理转写。 Modes pair a voice model with optional AI processing. Voice to Text skips AI processing; other modes can reshape the transcript. | 官方介绍自动去除语气词与重复、处理口述自我修正，并整理列表和段落。 Describes automatic filler and repetition removal, spoken self-correction, and formatting for lists and structured text. |
| 词汇与设置Vocabulary and controls | 手动保存人名和项目名作为识别提示，不等同于自动学习或正确率保证。iPhone 云端识别需要登录和网络。 Manually save names and project terms as recognition hints, not automatic learning or an accuracy guarantee. iPhone cloud transcription needs sign-in and internet. | 提供名字和常用词的个人词典；设置与功能因平台而异，转写需要网络。 Offers a dictionary for names and recurring terms. Controls differ by platform; transcription requires internet. | 文档介绍按模式选择本地或云端语音模型，语言支持取决于模型。请另行核对你的平台是否具备相同选项。 Documents local or cloud voice models per mode, with model-dependent language support. Check which options exist on your platform. | 官方介绍手动或自动添加的个人词典，以及混合语言输入。这些是产品说明，不是我们的独立测试结果。 Describes manual or automatic dictionary entries and mixed-language input. These are product descriptions, not our independent test results. |
官方来源：Official sources: Echo, Wispr Flow, Superwhisper modes, Typeless.
Try the free message test pack: check names, dates, and meaning before you send. 试用免费工作消息测试包：发送前检查人名、日期和原意是否保留。 合成例句，无需注册或上传录音；不是性能测评。 Synthetic examples. No signup or recording upload. Not a benchmark.
定价 Pricing
免费开始，需要更高强度时再升级。 Start free. Upgrade when Echo becomes daily infrastructure.
免费版 Free
$0/mo
日常输入、试用与自带 API Key 场景。 Daily dictation, testing, and bring-your-own-key workflows.
- 实时流式转录Real-time streaming
- AI 自动编辑与口述AI auto-editing
- 个人词典Personal dictionary
- 每周 5,000 字5,000 words per week
Pro
Echo Pro
$9.99/mo
7 天免费试用，解锁高强度输入。 7-day free trial for heavy daily use.
- 无限字数Unlimited words
- Smart Verify 交叉验证Smart Verify cross-engine checks
- 优先 Finalize 增强 ASRPriority finalize with enhanced ASR
- 高级 AI 润色预设Advanced AI polish presets
- 云同步与优先支持Cloud sync and priority support
| 功能 Feature | 免费版 Free | Pro |
|---|---|---|
| 实时流式转录 Real-time streaming | 已包含 Included | 已包含 Included |
| AI 自动编辑 AI auto-editing | 已包含 Included | 已包含 Included |
| 个人词典 Personal dictionary | 已包含 Included | 已包含 Included |
| 语言支持 Languages | 100+ | 100+ |
| 字数额度 Word quota | 5,000/周 5,000/wk | 无限 Unlimited |
| Smart Verify | — | 已包含 Included |
| 优先 Finalize Priority finalize | — | 已包含 Included |
| 高级 AI 润色预设 Advanced AI polish presets | — | 已包含 Included |
| 云同步与备份 Cloud sync and backup | — | 已包含 Included |
常见问题 FAQ
下载前，把关键问题讲清楚。 Key answers before the download path.
Echo 是什么？ What is Echo Voice Keyboard?
Echo 是一款面向 iOS 和 macOS 的 AI 语音输入层。它把实时转写、过程中修错、个人词典和 AI 润色放进同一个输入回路，让你边说边看、停下成稿、需要时再精修。 Echo is an AI voice input layer for iOS and macOS. It combines live transcription, in-stream repair, personal dictionary, and AI polish so you can review while speaking, pause into final text, and refine only when needed.
Echo 是免费的吗？ Is Echo free to use?
免费版包含每周 5,000 字、实时流式转写、AI 自动编辑和个人词典。Pro 版解锁无限字数、Smart Verify、优先 Finalize 和高级润色预设。 The free tier includes 5,000 words per week, real-time streaming, AI auto-editing, and personal dictionary. Pro unlocks unlimited words, Smart Verify, priority finalize, and advanced polish presets.
和 Apple 听写有什么区别？ How is Echo different from Apple Dictation?
Apple 听写更接近原始转录；Echo 的重点是边说边看到可用文本，并在转写过程中根据上下文修错。停下后再快速定稿，必要时才进入 Polish、Format 或 Translate。 Apple Dictation is closer to raw transcription. Echo focuses on usable text while you speak, context repair during transcription, a fast final after pause, and optional Polish, Format, or Translate when needed.
支持中英文混合输入吗？ Does Echo support Chinese-English mix?
支持。Echo 面向中英文代码切换做了专门体验设计，并支持 100+ 语言的语音转文字和翻译流程。 Yes. Echo is designed for Chinese-English code switching and supports 100+ languages across transcription and translation workflows.
Quick 会做 AI 润色吗？ Does Quick use AI polish?
不会。Quick 只做语音转写，不运行 LLM 改写。需要 AI 润色时可选择 Polish；格式化和翻译也按需启用，不属于 Quick。 No. Quick is transcription only, without LLM rewriting. Choose Polish when you want AI editing. Optional formatting and translation are separate from Quick.
需要自带 API Key 吗？ Do I need my own API key?
在 iPhone 上，登录后即可使用 Echo 提供的云端语音识别，无需自带 API Key，但需要网络连接。BYOK 是使用个人供应商账户时的可选配置，仅适用于支持的流程。 On iPhone, sign in to use Echo-provided cloud speech recognition without your own API key. An internet connection is required. BYOK is optional for supported workflows using your own provider account.
语音数据安全吗？ Is my voice data private?
Echo 以本地优先为设计原则：历史、个人词典和上下文优先留在设备上；API Key 使用系统安全存储。具体识别与修正路径取决于你选择的引擎和模式。 Echo is designed local-first: history, dictionary, and context stay on-device first, and API keys use secure system storage. Recognition and correction paths depend on the engine and mode you choose.
现在可用 Available now
让声音成为生产级文本。 Make speech production-grade text.
iOS 17+ 自定义键盘，macOS 14+ 菜单栏输入工具。Windows 与 Android 在路线图中。 iOS 17+ custom keyboard and macOS 14+ menu bar input tool. Windows and Android are on the roadmap.
免费下载。iPhone 用户登录后可使用 Echo 云端识别，无需自带 API Key。 Free to download. On iPhone, sign in for Echo cloud speech recognition. No personal API key needed.