🎧 IntegrityCheck — Forensic Audio Media Integrity Verifier

IntegrityCheck is an open-source script that forensically verifies whether YouTube/Invidious audio media has been tampered with.

It compares the audio file downloaded from your local machine against a copy downloaded inside a clean, ephemeral Microsoft Azure VM using the same tools, same URL, and the same session cookies.

The result? Verifiable, forensic-level integrity reports based on checksum mismatches — entirely independent of perception or subjective hearing.

🧠 Abstract

In environments where audio integrity cannot be assumed — either due to targeted tampering, surveillance, or psychological harassment — it becomes essential to verify that the media we consume is unaltered.

IntegrityCheck addresses this by:

• Eliminating perception-based claims
• Focusing only on binary file comparison
• Running in a controlled environment

🔧 What the Script Does

1. Downloads the same audio media twice — once locally, once inside a clean Azure VM
2. Aligns the environment: Ensures the same yt-dlp, ffmpeg, and Python versions are used on both ends
3. Uses the same session cookie file (cookies.txt) in both environments
4. Runs multiple cryptographic checksums: md5, sha256, sha512, and compares:
   • Binary file content
   • File size
   • (Optionally) metadata and duration
5. Reports any mismatch as proof of tampering
6. Deletes the VM post-download to ensure the cleanroom remains uncompromised

This design guarantees the only variable between the two files is the environment (local vs VM), not the download method.

✅ Features

• ✅ Verifies file integrity using md5, sha256, and sha512
• ✅ Consistent tools and versions across environments
• ✅ Ignores metadata from checksum calculation
• ✅ Preserves evidence with timestamped .txt and .json reports
• ✅ Supports YouTube and Invidious links
• ✅ Automatically spins up and tears down Azure VM
• ✅ Archives audio + metadata locally for future evidence

🧪 Sample Use Case

If you suspect media tampering or audio injection (e.g., subtle distortions, whispering, drops of water, gusts of wind), perform these steps:

• set up an azure account and make sure you input your credentials into the script where it is indicated "azure_user" and "azure_password"
• access youtube and export your cookies into the root directory of this folder (named as "cookies.txt"), using a chrome extension for example "Get cookies.txt locally" works fine.
• then run:

```bash
python3 init.py "<video_url>"