<?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>Token Saving on VibePapa — Tech from the Boseong River</title>
    <link>https://boseonggang.com/tags/token-saving/</link>
    <description>Recent content in Token Saving on VibePapa — Tech from the Boseong River</description>
    <generator>Hugo</generator>
    <language>en-us</language>
    <lastBuildDate>Thu, 22 Oct 2026 07:50:08 +0900</lastBuildDate>
    <atom:link href="https://boseonggang.com/tags/token-saving/index.xml" rel="self" type="application/rss+xml" />
    <item>
      <title>I Marked Who Was Speaking by Ear, Then Graded the AI&#39;s Speaker Labels Against My Marks</title>
      <link>https://boseonggang.com/posts/grading-ai-speaker-labels-against-my-own-ears/</link>
      <pubDate>Thu, 22 Oct 2026 07:50:08 +0900</pubDate>
      <guid>https://boseonggang.com/posts/grading-ai-speaker-labels-against-my-own-ears/</guid>
      <description>My local speech-to-text pipeline labeled who was talking in a 39-minute audio file, and I wasn&amp;#39;t sure it was right. I marked one speaker&amp;#39;s turns by hand while listening, then compared. The AI was right 91.5% of the time, and a script did the math instead of a language model.</description>
    </item>
  </channel>
</rss>
