<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
<channel>
  <title>MATRIX · EMA Lightning</title>
  <link>https://derekwang2002.github.io/ai-intelligence/en/projects/ema-lightning/</link>
  <description>Material changes in AI models, agents, engineering and infrastructure, with primary sources and recommendations.</description>
  <language>en</language>
  <lastBuildDate>Mon, 02 Nov 2026 16:56:00 GMT</lastBuildDate>
  <atom:link href="https://derekwang2002.github.io/ai-intelligence/en/feeds/projects/ema-lightning.xml" rel="self" type="application/rss+xml"/>
  <item>
    <title>EMA Lightning releases a compact streaming Turkish speech model</title>
    <link>https://derekwang2002.github.io/ai-intelligence/en/events/ev-20261001-10/</link>
    <guid isPermaLink="false">ev-20261001-10-published</guid>
    <pubDate>Wed, 07 Oct 2026 22:04:39 GMT</pubDate>
    <category>New</category>
    <category>TRIAL</category>
    <description>&lt;p&gt;EMA Lightning releases Apache 2.0 weights and code for an 8.6M-parameter Turkish text-to-speech stack with streaming, CPU support and a shared GPU scheduler. The model card reports 0.92% WER on Freya-TR-Eval, 3.86 ms to first audio and 440x real-time generation on an RTX 4090. Training used about 1,000 hours of private single-speaker audio.&lt;/p&gt;&lt;p&gt;Recommendation: TRIAL&lt;/p&gt;&lt;p&gt;Sources: &lt;a href=&quot;https://huggingface.co/canberkkkkkk/ema-lightning&quot;&gt;EMA Lightning model card&lt;/a&gt; · &lt;a href=&quot;https://github.com/canberk7/ema-lightning&quot;&gt;canberk7/ema-lightning&lt;/a&gt;&lt;/p&gt;</description>
  </item>
</channel>
</rss>
