The future of Web MiniDisc Pro's encoders
Hello everyone!
As you might have noticed in the past couple of weeks, the remote encoding server used to encode LP / ATRAC3 / ATRAC3plus isn't as stable as it used to be. This is because it was necessary to put it behind a Cloudflare "wall", due to the influx of AI bots and scrapers. Unfortunately Cloudflare sometimes blocks legitimate connections too, causing instability.
Because of that, I am planning on merging in support for a new, community-made ATRAC encoder - atracdenc-rs, made by a member of the MD Wiki Discord. This encoder was built by reverse engineering the original Sony tool via an LLM.
I would like to preface this by saying that I personally am against using LLMs in my projects, and I am not planning on ever using them to write any kind of update to Web MiniDisc, but I believe that the encoder I linked produces audio of sufficient quality to have it be used instead of the remote encoder. This would make it so the whole application would become client-only without any dependencies on external servers, making it easier and faster to use for everyone.
If this encoder were to be merged with WebMiniDisc, it would be used in the exact same way as the old `atracdenc` encoder was used - its code would be compiled to WebAssembly, and the bridge between it and the core of the application would be written by me, manually.
I plan on leaving the option to use the remote encoder, since it can be self-hosted, even if the Wiki hosted one were to go down.
Please let me know what you think about this solution.
EDIT: Clarification on the authorship of the library in question.
EDIT 2: The same person has a second encoder, which has not been published yet, which is a higher-quality reverse engineering, therefore I've crossed out the link.