[Bug]: openclaw插件生成的课堂没有声音

[hustxujinkang]

Bug Description

https://open.maic.chat/classroom/RbAe3WFCKc

http://open.maic.chat/classroom/mEdGZUhLR6

试了两个都是这样

Steps to Reproduce

安装openclaw的openmaic插件，配置使用托管模式，输入了access code
引导生成课堂

Expected Behavior

课堂播放的时候应该有声音，实际没有

Actual Behavior

都是无声的

Deployment Method

Local development (npm run dev / pnpm dev / yarn dev)

Browser

No response

Operating System

No response

Relevant Logs / Screenshots

Additional Context

No response

[YizukiAme]

This is likely the same root cause as #5 / #12 / #25 — when browser-native-tts is the selected TTS provider, the playback engine currently has no Web Speech API integration, so classroom playback falls back to a silent reading timer.

PR #28 (fix: use browser speechSynthesis for playback when browser-native-tts is selected) adds speechSynthesis.speak() support directly in PlaybackEngine, which should fix this. Will verify once merged.

[doge2mars]

无声是因为你没有把声音成功调用出来，然后就生成课了，所以你得花钱调用TTS，我是把源码改了，直接调用了edge-tts，然后生成的课就有声音了。

[YizukiAme]

Update: PR #28 (now merged) fixes browser-native TTS preview in the homepage/settings UI — it was calling the server API which rejects browser-native-tts, and Firefox had a silent-success bug.

However, for classrooms generated via the hosted API (open.maic.chat) or OpenClaw plugin, the root cause is different: /api/generate-classroom doesn't currently pass TTS generation parameters, so the classroom JSON contains no audio data at all. This is tracked in #49.

As @doge2mars noted, you need a working TTS provider (like edge-tts or an API-backed provider) configured at generation time for the classroom to contain audio. browser-native-tts only works for real-time playback, not for pre-generating audio into the classroom file.

[peter-gz]

无声是因为你没有把声音成功调用出来，然后就生成课了，所以你得花钱调用TTS，我是把源码改了，直接调用了edge-tts，然后生成的课就有声音了。

我改了没生效，请问是在哪个文件里改？

[doge2mars]

无声是因为你没有把声音成功调用出来，然后就生成课了，所以你得花钱调用TTS，我是把源码改了，直接调用了edge-tts，然后生成的课就有声音了。

我改了没生效，请问是在哪个文件里改？

我是交给antigravity直接帮我完成的，用上了edge-tts 而且是免费的

[wyuc]

和 @YizukiAme 分析的一样，之前通过 API 生成课堂时没有走服务端 TTS 流程，生成出来的课堂里没有音频。PR #75 已经修复了这个问题，API 现在支持 enableTTS: true 来做服务端音频生成，hosted mode 也部署了新版本。

感谢 @YizukiAme 的排查和 @doge2mars 给其他用户提供的 workaround。如果还有问题欢迎反馈。