Are the patents officially expired?

It's not just the parameters, even picking the algorithm itself is in scope for encoding. The MPEG audio layer specification doesn't mention how you might choose to go about encoding PCM data at all, it _solely_ specifies how an encoded stream is to be converted back into PCM data by a correct decoder. I haven't stared closely at MP3, but it isn't rare for these lossy encodings to have lesser features which, at the time they were standardised, nobody really knew how to use well when encoding. They're basically a few bits gambled on future proofing. A correct decoder should work fine regardless of whether that feature ends up being incredibly important or is instead left to rot because no researchers can figure out how to make use of it.

> I would expect that patents necessary for a decoder would
> expire sooner than patents for an encoder

In general it is exactly the opposite. In an encoder, you can pick and choose what you actually use. In a decoder, you have to be able to play anything someone might have chosen to encode. I believe it has been possible to write a not-crippled-so-much-as-to-be-useless MP3 encoder since 2013.

The difficulty isn't the complexity of the processing you have to do, it's the fact that you must comply with the standard. I.e., there is little to no value in these patents from the technology or the invention. All of the value is from the network effects of other people using them.