Currently, bytes whose 0xff mask is larger than 0xf0 are just ignored by the Buffers.readModifiedUtf8* methods.
Start treating those bytes by appending a replacement character to the string builder.