Experimental LSPosed module, scoped only to com.openai.chatgpt.

v1.2 tries to make ChatGPT voice output behave more like ordinary speaker/media output instead of a VoIP/phone stream:

• force BUILTIN_SPEAKER
• replace MODE_IN_COMMUNICATION with MODE_NORMAL
• replace USAGE_VOICE_COMMUNICATION with USAGE_MEDIA

Build: Build -> Assemble Project

APK: app/build/outputs/apk/debug/app-debug.apk

Test:

1. Disable old ChatGPT Speaker modules in LSPosed.
2. Install v1.2.
3. Enable v1.2 only for ChatGPT.
4. Reboot.
5. Test Voice Mode.

LSPosed log markers: MODE_IN_COMMUNICATION -> MODE_NORMAL USAGE_VOICE_COMMUNICATION -> USAGE_MEDIA AudioAttributes usage= forcing BUILTIN_SPEAKER

This is experimental. If it does not change the volume, the next likely layer is HyperOS AudioPolicy/AudioFlinger rather than another app-level hook.