Tracking: 0→50 stars 阶段真实试用反馈收集

[botaruibo]

这个 issue 用来追踪 MyVoiceTyping 在 0→50 stars 阶段最需要收集的真实反馈。

目标不是一次性堆功能，而是让 macOS 用户能顺利完成：下载 → 授权 → 录音 → 转写/润写 → 粘贴 → 反馈。

反馈收集目标

• 收集 10 条以上真实 macOS 试用反馈
• 至少覆盖 Apple Silicon M1 / M2 / M3 / M4 中的 2 种芯片
• 至少覆盖 3 类场景：聊天回复、文档/需求、AI Coding 或 PR 说明
• 收集 5 条以上可公开的“原始口述大意 / 实际输出 / 期望输出”样例
• 收集首次模型下载失败或权限失败的真实日志/截图

优先观察的问题

• 首次启动和权限引导是否清楚
• 模型下载是否稳定，失败提示是否足够明确
• 转写错字、同音字、技术术语识别问题
• 标点、断句、分段是否自然
• 本地 GGUF 润写是否过度改变原意
• 结果是否能稳定粘贴到当前输入位置
• CPU / 内存占用是否影响常驻使用

反馈入口

• FAQ / Troubleshooting: https://github.com/botaruibo/MyVoiceTyping/blob/main/docs/FAQ.md
• Privacy / Data Safety: https://github.com/botaruibo/MyVoiceTyping/blob/main/docs/PRIVACY.md
• Roadmap: https://github.com/botaruibo/MyVoiceTyping/blob/main/ROADMAP.md
• Contributing: https://github.com/botaruibo/MyVoiceTyping/blob/main/CONTRIBUTING.md
• 体验招募 Discussion: 招募 5 位 macOS 用户体验 MyVoiceTyping，欢迎反馈中文语音识别与本地润写效果 #2

反馈模板

请尽量使用 GitHub Issue 模板：

• Bug report / 故障报告：适合应用启动、权限、模型下载、录音、粘贴等问题
• 体验反馈 / User feedback：适合转写质量、润写质量和使用场景反馈

请不要上传包含公司机密、个人隐私、Token、密码或敏感内容的音频/文本。

[botaruibo]

Update: I have added a few onboarding pages so new testers can understand the project before installing it:

• Demo / 使用效果示例: https://github.com/botaruibo/MyVoiceTyping/blob/main/docs/DEMO.md
• Latest release notes: https://github.com/botaruibo/MyVoiceTyping/releases/tag/release-0.02
• Model usage / 模型使用说明: https://github.com/botaruibo/MyVoiceTyping/blob/main/docs/MODEL.md
• Privacy / Data Safety: https://github.com/botaruibo/MyVoiceTyping/blob/main/docs/PRIVACY.md
• Alternatives / Comparison: https://github.com/botaruibo/MyVoiceTyping/blob/main/docs/ALTERNATIVES.md

If you are trying MyVoiceTyping for the first time, I recommend starting with the Demo page first. It shows the intended flow: spoken Chinese → rough ASR text → punctuation / correction / light polishing → final text pasted into the current input field.

Most useful feedback right now:

• Does the app start correctly on your macOS version?
• Are microphone / accessibility / input monitoring permissions easy to configure?
• Is Chinese or Chinese-English mixed input accurate enough?
• Does the local polishing model keep your meaning, or does it over-rewrite?
• Which apps fail to receive pasted text?

Please avoid sharing sensitive audio or private work text. A short non-sensitive example is enough.

[botaruibo]

Update: for the current 0→50 stars stage, I am looking for 10 real macOS testers rather than broad promotion.

If you are willing to try MyVoiceTyping, the most useful feedback is:

• macOS version and Apple Silicon chip;
• whether the app starts correctly;
• whether microphone / accessibility / input monitoring permissions are clear;
• whether Chinese or Chinese-English mixed input works for your real use case;
• whether the local polishing model preserves your meaning or over-rewrites;
• whether the final text can be pasted into the app you are using;
• one non-sensitive example of: spoken intent → actual output → expected output.

Good test scenarios:

• chat replies;
• writing documents or requirements;
• AI Coding prompts / bug reproduction / PR notes;
• privacy-sensitive local input where cloud dictation feels uncomfortable.

Please do not share private audio, internal code, customer data, or sensitive work text. A short synthetic example is enough.

If you want Linux support, please use the separate demand-collection issue: #4

[botaruibo]

Update: I opened a separate feedback thread for AI Coding voice prompt examples:

#5

If your main use case is speaking bug reproduction steps, refactor instructions, PR notes, or prompts for Claude Code / Qwen Code / opencode / Cursor / Copilot Chat, please leave non-sensitive examples there. The most useful format is: spoken intent → actual output → expected output.

[botaruibo]

补充一个 30 秒试用任务，方便第一次点进来的朋友快速判断 MyVoiceTyping 是否适合自己：

1. 打开任意输入框，比如备忘录、浏览器文本框、GitHub Issue、Cursor/Claude Code 输入框。
2. 按住快捷键，说下面这句话：

帮我检查一下登录页面，用户点击发送验证码以后按钮应该进入倒计时，如果接口报错需要恢复按钮并显示错误信息。

3. 看最终粘贴出来的文本是否满足这几个点：

• “登录”“验证码”“倒计时”这些词是否正确；
• 是否自动补了标点和断句；
• 是否保留了原意，没有擅自增加需求；
• 是否适合直接发给 AI Coding 工具；
• 整个过程是否符合你对本地隐私/数据安全的预期。

如果愿意反馈，直接贴非敏感结果即可：

设备 / macOS：
使用场景：普通输入 / AI Coding / IM / 其他
口述内容：
实际输出：
你期望的输出：
最大问题：错字 / 标点 / 过度润写 / 粘贴失败 / 权限 / 模型下载 / 其他

注意：请不要贴公司内部代码、客户信息、真实日志、token、私人聊天或任何敏感内容。

[botaruibo]

补充：MyVoiceTyping 的普通用户落地页已经上线，适合先快速判断是不是值得安装：

https://botaruibo.github.io/MyVoiceTyping/landing/

建议新的试用路径：

1. 先看 landing 页的核心流程和截图；
2. 到 Release 下载最新版： https://github.com/botaruibo/MyVoiceTyping/releases
3. 用一段无隐私中文或中英混合内容做 30 秒测试；
4. 把安装、权限、模型下载、转写、标点、润写、粘贴问题反馈到这个 issue。

如果它确实解决了你的中文输入问题，也欢迎顺手 Star 一下项目，方便后续跟进版本：
https://github.com/botaruibo/MyVoiceTyping

仍然提醒：请不要提交公司内部内容、私人录音、聊天记录等敏感样例。用无隐私的短句测试就足够。

[botaruibo]

补充一个新的转化/反馈入口：中文同类工具对比页已经加入仓库。

• 中文版： https://github.com/botaruibo/MyVoiceTyping/blob/main/docs/ALTERNATIVES.zh-CN.md
• English version: https://github.com/botaruibo/MyVoiceTyping/blob/main/docs/ALTERNATIVES.md
• 对应提交： 1755403

这个页面主要回答中文社区里最常见的问题：MyVoiceTyping 和 Typeless、闪电说、Typeoff、Wispr Flow、Handy、MacParakeet / OpenQuack 等工具到底怎么选。

页面会刻意保持克制：

• 不说 MyVoiceTyping 已经完全替代成熟商业产品；
• 不说 100% 隐私或绝对不上云；
• 明确 Typeless / 闪电说 / Typeoff 在产品成熟度、跨平台、商业服务、云端 AI 编辑等方面可能更完整；
• 强调 MyVoiceTyping 当前更适合在意开源、本地优先、0 费用、中文 / 中英混合、代码 / 模型 / 数据集可审查的人。

后续从 Typeless / 闪电说 / Typeoff 对比讨论来的用户，可以先看这个页面再决定是否试用；如果对比描述不准确，也欢迎直接在这个 issue 里指出。

[botaruibo]

Update: added a dedicated Demo assets page for people who want to quickly understand or share MyVoiceTyping:

https://github.com/botaruibo/MyVoiceTyping/blob/main/docs/DEMO_ASSETS.md

It now collects:

• one-line pitch in Chinese / English;
• 5 launch cards / screenshots;
• 60-second demo storyboard;
• copy blocks for general sharing, AI Coding prompt scenarios, and data-safety discussions;
• recommended links for users and developers;
• wording guardrails so we do not overclaim privacy or say it fully replaces Typeless / 闪电说.

For first-time testers, the suggested path is now:

1. open the landing page: https://botaruibo.github.io/MyVoiceTyping/landing/
2. skim Demo assets / screenshots;
3. download the latest release;
4. try one short Chinese or Chinese-English prompt;
5. report startup, permission, model download, transcription, polishing, or paste issues here.

The most useful feedback is still real examples: rough spoken intent → actual output → expected output. Please remove any company secrets, private messages, tokens, or personal data before posting.

[botaruibo]

Update: added a dedicated macOS Quickstart guide for first-time testers:

https://github.com/botaruibo/MyVoiceTyping/blob/main/docs/QUICKSTART.zh-CN.md

It gives a short path for new users:

1. download the latest release;
2. grant the required macOS permissions: microphone, accessibility, input monitoring;
3. let the app download local models on first use if needed;
4. run a 30-second Chinese test prompt;
5. optionally run an AI Coding prompt test;
6. report issues with a structured template.

The goal is to reduce early drop-off from macOS permissions, model download confusion, and paste/focus issues.

For feedback, the most useful format is still:

• device / macOS version / chip;
• usage scenario;
• rough spoken intent;
• actual output;
• expected output;
• issue type: startup, permission, model download, recording, paste, typo, punctuation, over-polishing, technical terms, etc.

Please remove company secrets, private messages, tokens, passwords, internal code, and personal data before posting examples.

[botaruibo]

Update: I added a dedicated 30-second trial tasks page so first-time testers do not need to invent their own test text:

https://github.com/botaruibo/MyVoiceTyping/blob/main/docs/TRIAL_TASKS.zh-CN.md

It includes 6 synthetic, non-sensitive tasks:

• AI Coding prompt
• work message / email
• Chinese-English technical terms
• bug reproduction note
• privacy-sensitive input safety test
• self-evaluation / local personalization sample

For this 0→50 stars stage, the most useful feedback is still simple and concrete: pick one task, run it in Notes or your target input box, then share a sanitized spoken intent → actual output → expected output example.

Please do not post company secrets, private chat text, tokens, customer data, or internal code. Synthetic examples are enough.

[botaruibo]

补充一个反馈入口：如果试用时遇到的是“可公开、已脱敏”的 ASR 后处理样例，也可以同步到 Dataset 仓库的样例征集 issue：

botaruibo/MyVoiceTyping-Dataset#1

适合放在那里的样例包括：

1. 中文音近错词；
2. 中英混合里的技术词、库名、文件名；
3. 标点和断句不自然；
4. 润写过度、改变原意；
5. 用户手动改写前后的对照。

建议格式：

场景：
ASR 原文：
期望文本：
哪些词不能改：
备注：

隐私提醒：不要提交公司机密、客户信息、token、密码、私人聊天记录或未脱敏日志。这个入口主要用来积累可公开的失败样例，方便后续改进本地润写模型和 self-evaluation 流程。

[botaruibo]

Update: 为了让从社区讨论点进来的用户更容易判断是否值得试用 / Star，我补齐了两类承接材料：

• FAQ 里的「从 Typeless / 闪电说 / Typeoff 讨论过来的」快速判断路径：
  https://github.com/botaruibo/MyVoiceTyping/blob/main/docs/FAQ.md
• Press kit / 投稿包，方便周刊、产品目录、社区推荐和转发时直接引用：
  https://github.com/botaruibo/MyVoiceTyping/blob/main/docs/PRESS_KIT.md

如果你只是想快速判断是否适合自己，建议先用 30 秒试用任务跑一遍虚构句子，不需要一上来就输入真实公司内容：
https://github.com/botaruibo/MyVoiceTyping/blob/main/docs/TRIAL_TASKS.zh-CN.md

如果方向对你有用，欢迎 Star 关注后续 Demo、License 和中文 / 中英混合输入体验改进；如果暂时不适合，也欢迎直接在这个 issue 里说出卡点。