MohammadThabetHassan/VoiceGuard ? reverse-engineered prompt

Build me a voice security web app that can tell if an audio clip is real or a deepfake, and show a clear result in real time. I want users to be able to upload a recording, use their microphone live, and get a simple verdict with an explanation of what parts of the audio influenced the decision.

It should also have a voice generation area where an admin can create speech from text, optionally clone a voice from a reference clip, and automatically watermark anything generated so it can be checked later. Add a verification screen that can confirm whether a clip was produced by the app, plus a basic forensic report export with a chain of custody style record.

Make the interface clean and modern, with a working backend API and a React frontend. If you need to look up current docs for the audio and streaming parts, go ahead and do that.