闪念Spark vs Thought_Box

闪念Spark 只做一件事：把你此刻想到的那句话，用最少的动作留下来。 "Spark Notes" does only one thing: catching your thoughts the moment they occur. 没有标签，没有文件夹，没有每日回顾，没有统计图表。打开就是一个输入框、 一个录音键和一条时间流。 按住就录，松手就存 走在路上一个念头冒出来——从锁屏小组件或控制中心按一下，进来的时候已经在录了。 上滑锁定可以一直说下去，锁屏之后录音继续，锁屏上看得到时长，也能暂停和结束。 说完，文字就在屏幕上 转写全程在你的 iPhone 上完成，不联网。中文专名（圣经书卷、神学词汇）做过识别优化。 支持 Apple Watch：抬腕说一句，自动送到 iPhone。 录音永远不会丢 录音先落盘，再尝试转写。转写失败、模型没装好、权限被拒、App 被划掉——宁可留下 一条只有录音没有文字的笔记，也绝不丢掉你说过的话。 想写字的时候 右下角一个加号，在列表最前面弹出一行空白，写完点别处就收起。停笔一秒自动保存， 直接按电源键锁屏也不丢字。 搜索 一个搜索框，全文匹配。 导出到 Obsidian（免费） 选一个「文件」App 里能看到的文件夹——iCloud 云盘、坚果云、Dropbox 都可以， 只要它在你的 Obsidian 库里。每条笔记自动写成一份 Markdown。你在 Obsidian 里 手动改过的文件不会被覆盖。 AI 润饰（App 内购买解锁，一次性买断，非订阅） 端上转写十句里总有一两个词不对，那是机器听错，不是你想错。解锁之后可以填自己的 服务地址、模型名和 API Key（OpenAI 兼容格式，DeepSeek、通义、智谱、月之暗面都行， 也可以是你自己架的 Ollama），转写完的文字交给它加标点、断句、改同音错字、去口水词。 用的是你自己的额度，Spark 不代收订阅、不做中转，也不经手你的内容。录音永远不出 这台手机，只发送转写出来的文字，而且要你亲手点过一次确认才开始发。原始转写永久 保留，播放条那里随时能切回去看，导出的 Markdown 里也带着原文。 AI 只做这一件事——不总结、不扩写、不提炼观点、不加你没说过的话。 不做的事 不从微信、网页或任何外部渠道导入内容。没有账号、没有登录、没有广告、 没有任何分析埋点。

Give your passing ideas somewhere to grow. Thought Box brings everyday capture, playful rediscovery, and deeper reflection into one personal library. CAPTURE IN YOUR OWN WAY Type a thought, record your voice, read text with the camera, or import a photo or file. Bring in supported content from Apple Notes through sharing or import. Use the app, a widget, or a Shortcut configured for a supported Action button or Back Tap. REDISCOVER WHAT YOU SAVED Browse your records with Cover Flow and an iPod-inspired click wheel. Let Home Screen widgets bring saved passages back into view. Keep Markdown notes alongside their source files. THINK DEEPER With your own supported AI provider API key, Inspired Talk helps you discuss a record, consider another perspective, and develop the thought behind it. Ask searches related records and builds answers with source citations you can open and check. YOUR LIBRARY, YOUR CHOICE Capture, browsing, and local search work without an AI key. Choose local storage or an iCloud Drive library; synchronization depends on your Apple account, device, network, and iCloud availability. Speech transcription uses supported on-device recognition and may require a language-model download. Optional AI features send the disclosed text to your selected provider after you grant permission. Cloud Read Aloud offers Qwen (DashScope Beijing) and GLM (BigModel mainland China), with a separate voice API key and separate text-sharing permission; on-device speech works without an AI key. You supply the provider API key and are responsible for any provider charges. Provider availability, limits, and data handling vary by service. AI responses can be mistaken; check their sources. Available in English, Simplified Chinese, and Traditional Chinese. This version offers supported domestic text AI services. Before sending content, the app identifies the recipient, data and purpose and asks for your explicit permission. You can turn off text sharing at any time in Settings or on the relevant feature page. Cloud speech uses independent sharing permission.