Music Recommendation System for a Workout Playlist

Music Recommendation System (the "Suggested Playlists/Artists/Songs" you get on Apple Music/Spotify) but strictly based on acoustic features: compare acoustic features/spectral properties such as tempo/BPM (beats per minute) of your most played songs in a workout playlist to those of all available songs to recommend songs/playlist with similar acoustic features.

I know the music streaming services use algorithms that analyze audio and recognize various characteristics of the music such as genre (abstract) or type of instruments but not sure there is one that specifically caters to the tempo. Alternatively, tempo could be swapped out for a different feature such as pitch or timbre.