These are chat archives for FreeCodeCamp/DataScience

26th
Jan 2019
Alice Jiang
@becausealice2
Jan 26 01:43
Inspired by last night's Data & Music shindig, I've been digging around the Spotify API again and I feel pretty confident that it could be extremely useful to my project with the billboard charts, and have some ideas for another handful of small projects... but I've run into the issue of I'm not confident the data returned in the analyses of a track are cross-cultural. For example, you can request the audio features of a track, and one of the keys returned in the features object is "valence" with a measure from 0.0 to 1.0 describing the musical positiveness conveyed by a track (TL;DR: valence is a binary measure of emotions, positive like happiness and excitement, or negative like sad or angry) which presents a problem internationally as musical positivity and musical negativity sound different across cultures, an issue which in large part goes hand in hand with the returned key-value pair for "mode" which describes a songs modality (think major/minor scales). Not only are western major and minor scales not universally distinguishable as happy and sad, most non-western cultures don't even us the same 12-tone scales that we do.
It might just because my programming background has largely been centered around accessibility and internationalization, but how would you go about addressing internationalization regarding a subject that has extremely different rules depending on the geographic region?
Alice Jiang
@becausealice2
Jan 26 02:12
Also, When you open up browser console at the Spotify Web API reference page The site devs have left a secret message x)
Philip Durbin
@pdurbin
Jan 26 03:00
That's so cool that you went. My wife had a work event and I stayed home with the kids.
Alice Jiang
@becausealice2
Jan 26 03:37
It was fun! They didn't go too in depth, Spotify's legal team asked them to stick to black box explanations, but they talked about some of the projects they're working on and interesting problems they face
My favorite was when talking about their voice system, they have to account for contextual requests for music and returning someting appropriate, so instead of asking specifically for a song or artist or genre, people are asking Spotify to "play guacamole making music" and they get the challenge of how to return a single correct song on the first try lol
she says they've got the system to where it knows how to respond to "play something lit" which is hysterical, tbh
Philip Durbin
@pdurbin
Jan 26 03:45
:)