<?xml version="1.0" encoding="UTF-8"?><oembed><type>video</type><version>1.0</version><html>&lt;iframe src=&quot;https://www.loom.com/embed/1cdf1468090f438a859d04430c0f1ca9&quot; frameborder=&quot;0&quot; width=&quot;2626&quot; height=&quot;1969&quot; webkitallowfullscreen mozallowfullscreen allowfullscreen&gt;&lt;/iframe&gt;</html><height>1969</height><width>2626</width><provider_name>Loom</provider_name><provider_url>https://www.loom.com</provider_url><thumbnail_height>1969</thumbnail_height><thumbnail_width>2626</thumbnail_width><thumbnail_url>https://cdn.loom.com/sessions/thumbnails/1cdf1468090f438a859d04430c0f1ca9-77b39470244e6ad4.gif</thumbnail_url><duration>83.968</duration><title>Overview of the New TTS Engine and Export Pipeline</title><description>In this video, I introduce our new standalone node that establishes a shared data model, with TourStop as the central type. We&apos;ve replaced the old TTS stack with a new TTS engine that utilizes cloud APIs to generate mp3 files. I also discuss the FromEnvironmentFactory&apos;s ability to auto-detect between OpenAI and Alep, and how the synthesize method incorporates retry logic with exponential backoff. It&apos;s important to note that our execute method manages concurrency to prevent exceeding TTS API rate limits. Please review the implementation details and let me know if you have any questions or feedback.</description></oembed>