“Hey, Spiridon”唤醒词的开发者在训练模型时遇到了重大挑战。最初使用 Piper 和 openWakeWord 笔记本的尝试因依赖问题和声音多样性不足而受阻。随后，开发者转向 ElevenLabs，发现其生成的首批样本更加自然和通用。 AI
影响 这是唤醒词生成领域的一个小众工具改进，对整个行业影响甚微。
排序理由 开发者个人项目工具切换，并非重大的行业发布或产品。
在 Mastodon — fosstodon.org 阅读 →
AI 生成摘要 · Google Gemini · 来自 2 个来源。 我们如何撰写摘要 →
Goal: generate positive and hard-negative "Hey, Spiridon" wake-word samples. The sample generation with Piper worked, but training the model using the openWakeWord notebook turned into hours of untangling dependency issues. Combining that with Piper's lack of vocal variety, I aba…
Current goal: generate positive and hard-negative wake-word samples for "Hey, Spiridon". While working with the openWakeWord training notebook and Piper, I spent hours untangling dependency issues before abandoning that workflow entirely. Piper also wasn't giving me the pronuncia…