reddit · SideProject · 09-26 07:50 48 分 ← 返回列表

I got dictation from 4.1% to 2.4% word error rate and names from 85% to 100%. Would anyone actually use this?

I built a macOS dictation tool over the last few months because every one I tried mangled Indian names. Hold right Option, talk, release, and cleaned up text lands at the cursor in whatever app you are in. Whisper transcribes locally, then an LLM pass fixes it up. I am at the point where it works well enough that I use it all day, and I genuinely cannot tell whether that means anything to anyone else. So: numbers first, then the full feature list, then the questions I actually want answered

五维评分

付费意愿

需求真实度

个人可落地

竞争空间

新鲜度

AI 分析

痛点：现有 macOS 听写工具对印度/非英文人名的识别极差，转写出来名字错了整段文本就没法用，尤其是需要频繁输入同事、客户、公司名的场景。

目标用户：以英语口述但常涉及南亚/非英语人名的知识工作者、开发者、医生、律师、客服、留学生等 macOS 重度用户。

证据：every one I tried mangled Indian names ... a transcript with the right words and the wrong name is still unusable ... accented proper nouns went from 85 percent correct to 100 percent ... I genuinely cannot tell whether that means anything to anyone else.

变现：按月订阅（$8–15/月）卖本地优先的听写工具，后续可加行业/公司自定义词库包、团队词库共享、企业私有化部署等增值层。

MVP：直接把现有原型打包成付费 beta：只做「按住右 Option 说话→光标处出字」一条链路，主打自定义人名/公司名词库与口音纠正，用 Waitlist + 一次买断/早鸟订阅验证付费，不做多平台。

竞品：Wispr Flow、Superwhisper、MacWhisper、VoiceInk、Aqua Voice

摘要：一位开发者用本地 Whisper + LLM 清理做了 macOS 听写工具，把词错率从 4.1% 降到 2.4%，带口音的专有名词准确率从 85% 提到 100%，补齐了现有工具「名字总听错」的短板。但他明确说自己不确定是否有人需要，全文没有任何付费用户或收入数据，属于典型的自证型需求。听写赛道已有 Wispr Flow 等成熟付费竞品，差异化点（口音人名/自定义词库）是真实缺口，但付费意愿尚未被验证。