added 10 second limit in description

simplify segment detection to match paper's implementation

Switch to framewise sound-event localization with AudioLabel

Ignore HTS-AT-repo/ reference clone

Deploy HTS-AT AudioSet tagging model as HARP Space

initial commit