reddit · SideProject · 11-13 01:50 44 分 ← 返回列表

I got tired of proofreading everything I said out loud, so I built my own dictation app

I do most of my writing out loud. Messages, emails, AI prompting, notes to myself etc. Half of it happens at night, talking quietly into my phone because others are asleep in the next room and I do not want to wake them just to answer a message. For a long time I used the big and free dictation apps. The cleanup was fine until I kept catching sentences that were not what I said. Nothing dramatic, just small changes in meaning here and there that I could not send without checking and fixing.

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：现有免费听写工具会悄悄改动语义、把中英混合句子译成奇怪的英文，用户仍得逐句校对，反而更费时

目标用户：大量用语音写作、常在夜间小声口述、且经常中英或多语言混说的重度用户

证据：the tool was supposed to save me time. If I have to fix what it writes, I might as well type it myself. / It still rewords things with intention and care... It just does not get to change the meaning.

变现：订阅制（按月或按年），免费额度 + 付费解锁多语言混说、本地离线模型、自定义词典

MVP：基于 Whisper 等开源 ASR，做移动端快速口述 → 仅做不改语义的清理（去口水词、标点、断句）→ 保留多语言混说不翻译，2 周可做出可用 MVP

竞品：系统自带听写（iOS/Android）、Whisper 及本地离线听写 App、Superwhisper、MacWhisper、Wispr Flow

摘要：作者因免费听写 App 常悄悄改动语义、多语言混说被乱译而需反复校对，于是自建 Olomo（olomo.app），强调“只清理不改变意思”，自己每天用几个月后基本无需修改。但帖子无他人附和、无付费意愿或收入数据，属自推产品，商业验证仍不足。