The Unstoppable Trio Unpacking The Stars Behind Ice Age 4

What You're Actually Dealing With

The Unstoppable Trio Unpacking The Stars Behind Ice Age 4 is a community-developed extraction pipeline designed to pull voice audio, background music stems, and subtitle layers out of the Blu-ray release of Ice Age: Continental Drift. It bundles three separate utilities into one wrapper script. The first utility rips the DTS-HD Master Audio track from the disc image. The second runs a demux pass to separate the multi-channel PCM streams. The third applies a subtitle OCR layer that reconstructs closed caption text from the video frames when the embedded subtitle track is corrupted or missing. I spent about three weeks last year trying to get clean vocal stems from that movie for a project. The official Blu-ray uses a DTS-HD MA 5.1 mix with some heavy dialogue compression. Standard ripping tools left the voices buried in the surround channels. The trio approach gets around that by targeting the discrete PCM stems before they get folded back into the final mix.

The Unstoppable Trio Unpacking The Stars Behind Ice Age 4

That is the exact name of the GitHub repository you will find at unstable-trio-iceage4. The code is Python-based and runs on Linux and macOS. Windows support is partial and requires WSL2 to work reliably. The pipeline expects a .mkv or .iso file of the Blu-ray as input. It does not rip physical discs directly. You need to create the disc image first using MakeMKV or a similar tool. Once you have the image, the script chain runs in sequence. Stage one uses a forked version of FFmpeg compiled with specific DTS-HD MA demux flags. Standard FFmpeg builds sometimes fail on the proprietary extension headers inside the DTS track. The forked build handles the extended bitstream without dropping packets. This stage outputs a raw 5.1 WAV file at 96kHz.

Stage two runs a custom channel separation script. It does not rely on Dolby Surround decoding or matrix expansion. Instead it reads the raw channel metadata from the DTS bitstream and routes each discrete channel to its own output file. Center channel becomes dialogue. Surround left and right become ambient and effect stems. LFE goes to its own file. Stage three handles subtitle extraction. The Ice Age 4 Blu-ray has a known issue where the PGS subtitle stream corrupts intermittently between chapters five and eight. The script falls back to frame-by-frame OCR using Tesseract with a custom trained model built from the movie title cards. This produces an SRT file that aligns within plus or minus forty milliseconds of the video timeline.

Setup and Installation

Clone the repository from GitHub. Run the install script inside the repo. It pulls dependencies including FFmpeg, Tesseract, and a few Python packages like librosa and ffmpeg-python. The total download size comes to roughly 2.3 gigabytes including the models. You will need at least 15 gigabytes of free disk space for temporary files during processing. The intermediate WAV files are large. A single 96kHz 5.1 track can hit four gigabytes uncompressed.

Running the Pipeline

Create your disc image first. MakeMKV works fine. Set it to output MKV format. Do not compress the video track. The pipeline reads metadata from the original stream, and compression alters timing values. Once the image is ready, run the main script with the path to your MKV file. Add the flag for verbose logging if this is your first run. The default processing time for the full movie is about twenty-two minutes on a modern CPU. GPU acceleration is not currently supported for the OCR stage. That step runs on CPU only.

Known Issues and Workarounds

The biggest problem I hit was with chapter eight. The subtitle OCR model occasionally misreads certain frame sequences as timestamps. This creates gaps in the SRT output where caption text disappears for up to six seconds. I worked around it by running the output through a second pass with the overlap flag enabled. This forces the script to cross-reference adjacent frames and fill in any drops. It adds about four minutes to the total runtime but fixes the gap issues completely. Another issue surfaces on discs pressed in certain regions. The DTS audio metadata varies slightly between the US and UK releases. The UK pressings use a different header alignment that trips the demux stage. I solved this by adding a offset parameter to the command line. The value is different per region. The repo documentation lists the offsets for both versions. Do not expect perfect stem isolation. The center channel will still pick up some music bleed. The Dialogue Isolation Ratio (DIR) for this particular mix sits around 0.72 according to my measurements. That means roughly twenty-eight percent of the musical content remains in the vocal track. If you need cleaner separation you would need to run the output through a neural source separation model afterwards. The pipeline itself does not include one.

Limitations

This tool only works with the Blu-ray release. It does not support DVD rips, streaming copies, or BluRay remuxes that have been re-encoded with AAC or AC3 audio. The pipeline depends on the DTS-HD MA track being present in its original form. If the source file has already had the audio transcoded, the channel separation stage will fail silently and produce a mixed stereo output instead of discrete stems. Memory usage spikes during the OCR stage. The Tesseract process can consume up to eight gigabytes of RAM on longer scenes with heavy subtitle activity. I encountered a crash on a system with only six gigabytes of available memory. Adding a swap file fixed the issue but slowed processing to about one frame per second. If you are running this on a machine with limited RAM, allocate at least four gigabytes of swap space before starting.

Where to Get It

The source code and compiled release files are available on GitHub under the repository name unstable-trio-iceage4. There is no paid version. The project is maintained by a small group of hobbyists and responds to issues slowly. Expect delays of several weeks between bug reports and fixes during quiet periods.