hn · Show HN · 10-29 17:48 53 分 ← 返回列表

Show HN: Overtone – Transcribe, correct and cut captions for any screen shape

原文链接 ↗ ·rvanderlinden

Hi HN, I'm Ruud, co-founder of Lont. We turn company emails into personalized videos. Video captions cost us manual work twice: fixing misheard words, then fixing layout, like lines ending on “the”. Overtone uses multiple models to listen to the audio: gpt-transcribe, Azure STT, and Elevenlabs Scribe. When they return conflicting results, Overtone sends small clips to the models again, and then lets an LLM decide what was actually said. The LLM is like a human doing a sanity check on the transcr

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：视频字幕转录错误多、不同屏幕比例下排版需手动修正，耗时且容易漏错。

目标用户：视频制作团队、营销团队（如把邮件做成个性化视频的公司）、内容创作者。

证据：Video captions cost us manual work twice: fixing misheard words, then fixing layout... Sure, the cost is higher because we use up to 3 different STT models + an LLM, but it still removes mistakes... For us it's worth it.

变现：托管SaaS订阅（按视频分钟数或席位收费）、转录纠错API、企业私有部署版。

MVP：基于开源Overtone封装一个网页：上传视频→多STT模型转录+LLM交叉纠错→按16:9/9:16规则切分字幕→导出SRT或烧录，2周可完成。

竞品：Descript、Veed、CapCut、Rev、Whisper-based caption tools

摘要：Lont团队因制作个性化视频时字幕修正耗时，推出开源工具Overtone，用gpt-transcribe、Azure STT、Elevenlabs Scribe三个模型加LLM交叉校验降低WER，并按屏幕比例规则切分字幕。作者称成本更高但值得，但帖子仅10分0评论，外部需求尚未验证。