<?xml version="1.0" encoding="utf-8" standalone="yes"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:content="http://purl.org/rss/1.0/modules/content/">
  <channel>
    <title>Transcript Correction on VibePapa — Tech from the Boseong River</title>
    <link>https://boseonggang.com/tags/transcript-correction/</link>
    <description>Recent content in Transcript Correction on VibePapa — Tech from the Boseong River</description>
    <generator>Hugo</generator>
    <language>en-us</language>
    <lastBuildDate>Thu, 22 Oct 2026 09:07:18 +0900</lastBuildDate>
    <atom:link href="https://boseonggang.com/tags/transcript-correction/index.xml" rel="self" type="application/rss+xml" />
    <item>
      <title>Keep What the Speaker Said: Checking a Sermon Transcript Against the Bible Without Letting the AI Rewrite It</title>
      <link>https://boseonggang.com/posts/keep-what-the-speaker-said-checking-a-transcript-against-the-bible/</link>
      <pubDate>Thu, 22 Oct 2026 09:07:18 +0900</pubDate>
      <guid>https://boseonggang.com/posts/keep-what-the-speaker-said-checking-a-transcript-against-the-bible/</guid>
      <description>I wanted an AI to compare a sermon transcript with the Bible and fix misheard names. Left alone, it rewrote the speaker&amp;#39;s own words into the modern translation. Here is the gate that stopped it, the bug where the AI moved text up one line, and the cost in dollars.</description>
    </item>
    <item>
      <title>Local Qwen vs Gemini for Fixing a Transcript: 50 Minutes Free vs 2.3 Minutes for 9 Cents</title>
      <link>https://boseonggang.com/posts/local-qwen-vs-gemini-correcting-a-sermon-transcript/</link>
      <pubDate>Thu, 22 Oct 2026 08:31:33 +0900</pubDate>
      <guid>https://boseonggang.com/posts/local-qwen-vs-gemini-correcting-a-sermon-transcript/</guid>
      <description>I wanted to correct speech-to-text errors in a Korean sermon transcript without sending it to a cloud API, so I tried a local Qwen model. It took about 50 minutes and made new mistakes. Gemini took 2.3 minutes and cost $0.086. The comparison, with numbers and what each model got wrong.</description>
    </item>
  </channel>
</rss>
