For instance, seven live recordings of “Folsom Prison Blues” exist in the data set, and four studio recordings (Mercury Albums, Total Johnny Cash Sun Collection, All About the Blue Train, and I Walk the Line), giving us 28 possible matches to compare among Live to Studio.
This yielded multiple matches for a single song. Using the new binary indicator of live/studio, we matched tracks by track name and artist name. We removed tracks with “Commentary”, “Rehearsals”, “Intro”, “Movie Soundtrack”, or “Demos” from the analysis, and only kept artists who had at least 7 recorded live songs, per the previously created variable, as an indicator that live songs are representative of a live album recording, rather than a one off recording. We hard coded a binary indicator of live/studio by identifying tracks with the string “Live” or “Sessions” in the track name or in the album name. Raw SpotifyAPI data includes song level variables such as acoustics, speech levels, and duration of a song, and album level variables such as album name and year recorded. We culled this list of artists by selecting only artists who had charted on the Billboard Hot 100. We pulled the intersected list of artist names (about 10,950) from the Spotify API, and ended up with around 8,000 artists’ Spotify API data.
Our list of artists comes from Wikipedia’s list of live recorded albums, compared to the set of Pop/Rock artists who recorded at least one live recorded album by, as they appear on AllMusic. Alas, we might have missed your favorite artist, but hope that we struck some sort of comprehensive list, and encourage you to dig for that underground live version you can’t find here.
One of our conceptual challenges was figuring out how to get a comprehensive list of artists and albums, while still parsing down the data to create a functional tool. There are an estimated 10,000 artists who have recorded live albums in recording history. Then, by looking at the differences between live and studio songs using numerical metrics like energy (how intense or lively a track sounds), valence (a measurement of a song’s mood), and duration (to find those five-minute guitar solos), we found great live performances that we had missed. First, we found as many live albums as we could and matched them to their studio counterparts using the SpotifyAPI. In the time before we can go back to crowding Brooklyn Steel or catching an intimate set at Hole in the Wall, we decided to dive into the vast discography of live-recorded albums to try and satiate the desire to catch a show. What exactly do we miss? Is it the random banter outside of the lyrics we know by heart? A slow jam revived as an up-tempo frenzy? The spontaneous, five-minute guitar solos? A banger gone acoustic? Or the solidarity of strangers in the audience who love that band as much as you do? Maybe it’s the surrounding context of the show (Johnny Cash’s prison performances, Nirvana’s intimate NYC acoustic set, or LCD Soundsystem’s “final goodbye” Madison Square Garden show), that give us insight into what was going on with the audience, the artist, and the world, at the time of the performance. Of course, with modern streaming platforms, there’s more than enough recorded music to keep us safely locked down ad infinitum yet there remains a deficiency in our musical diets amidst the plague of cancelled shows and closed venues. It’s been a long time since we’ve seen live music.