←
hurf2,560 posts
hurf avatar

hurf

@hurfyd

cosine.club

◷ Joined August 2011▧ 507 Photos and videos
hurf avatar
hurf@hurfyd·Oct 29

hey sorry i havent replied to ur email about uploading tunes ive been busy working on a machine learning model to automatically ingest my inbox and score and rank them according to my taste so i can reply to emails faster

hurf avatar
hurf@hurfyd·Jan 27

who rememebrs julio bashmore battle for middle you remember that 1? julio bashmore battle for middle you who remembers it by julio bashmore battle for middle you

hurf avatar
hurf@hurfyd·18 Dec 2025

different to cosine.club which generates embeddings from mel spectrograms of the raw audio with metadata as supervision signal on the underlying model. also s/o to @_barneyhill for building a UMAP visualiser for cosine embeddings at barneyhill.com/pages/fieldst…

hurf avatar
hurf@hurfyd·18 Dec 2025

been having fun digging with this new discovery tool its a "spatial interface for non-linear exploration" aka a map of subgenres and styles in distinct islands you can navigate around and explore

Post attachment
hurf avatar
hurf@hurfyd·22 Oct 2025

made some updates recently have a look :) :) :) - 1.9 million tracks ingested now - u can save tracks to playlists - theres an API - filter on year, have/want, price - i made it green - thats it thank u every1 who sends nice messages and feedback cosine.club

hurf@hurfyd14 Apr 2024

building an audio similarity search engine.. given an input track it will find others that sound similar using vector embeddings from a machine learning model with an index of 1.1 million electronic music tracks

1:43
hurf avatar
hurf@hurfyd·11 Mar 2025

you can also click on the locations to zoom to that area (based on the locations labels/artists assign to their releases) or also start a new search using tags to find new stuff with those tags

hurf avatar
hurf@hurfyd·11 Mar 2025

been collaborating on a new music discovery tool using fan connections instead of algorithms you can give a bandcamp release and it will find everyone who bought it and show you what else they recently purchased on a map

0:45