<?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>重采样 on 技术博客 - 有条工具</title><link>/blog/tags/%E9%87%8D%E9%87%87%E6%A0%B7/</link><description>Recent content in 重采样 on 技术博客 - 有条工具</description><generator>Hugo -- 0.147.7</generator><language>zh-cn</language><copyright>2024-2025 有条工具技术博客</copyright><lastBuildDate>Wed, 11 Nov 2026 08:50:00 +0800</lastBuildDate><atom:link href="/blog/tags/%E9%87%8D%E9%87%87%E6%A0%B7/feed.xml" rel="self" type="application/rss+xml"/><item><title>音频变速的两条路：线性插值重采样，与 2048 采样窗的 WOLA 时间拉伸</title><link>/blog/articles/audio-time-stretch-wola/</link><pubDate>Wed, 11 Nov 2026 08:50:00 +0800</pubDate><guid>/blog/articles/audio-time-stretch-wola/</guid><description>1.5 倍速播 podcast，声音变尖了还是没变？取决于走哪条路：重采样是按倍距重读采样、音高随速度一起变，一行循环就能写完；变速不变调则要把音频切成 2048 采样的颗粒、加 Hann 窗重叠搬移、再按窗权归一化（WOLA）。这篇把两条路的数学、参数取舍与 4 倍重叠的窗和陷阱逐一拆开。</description></item></channel></rss>