📄 中文摘要
许多语音助手演示在短短30秒内看起来令人印象深刻，但当你试图将其视为真正的产品时，它们往往会崩溃。为了避免这种情况，作者构建了一个基于 ElevenLabs 的本地 Python 语音助手原型，旨在实现从实验到最小可行产品（MVP）的顺利过渡。文章提供了完整的代码和设置细节，并强调了在构建语音助手时需要关注的关键工程经验，以确保项目的可持续性和实用性。
从使用 ElevenLabs 和 Python 构建语音助手中获得的三条工程经验
📄 English Summary
Three engineering lessons from building a voice agent with ElevenLabs and Python
Many voice agent demos appear impressive for the first 30 seconds but tend to fall apart when treated as real products. To avoid this, the author created a local Python voice agent prototype using ElevenLabs, aiming for a smooth transition from experiment to minimum viable product (MVP). The article provides complete code and setup details, emphasizing key engineering lessons that are crucial for ensuring the sustainability and practicality of the voice agent project.