TIL: Take uncompressed audio with a sample peak not below -1.0 and not above 1.0, send it through an mp3 encoding/decoding round-trip, and it might come out with a sample peak below -1.0 and above 1.0 (!). This is (depending on encoding parameters) by design, and encoders expect decoders (or its consumers) to clamp/clip these peak overshoots themselves. Reality never ceases to surprise me with new intricacies where I didn't expect them. :)
Welcome to post.lurk.org, an instance for discussions around cultural freedom, experimental, new media art, net and computational culture, and things like that.