Example of using `calc_self_sim()` for self-similarity of song lyrics

lyric_self_similarity

library(genius)
library(tidytext)
library(tidyverse)


# get lyrics for album
since_96 <- genius_album("ALIX", "since 96")

# calculate self similarity by group (thank you dplyr v 0.8!!!)
self_sim_96 <- since_96 %>% 
  group_by(track_title, track_n) %>% 
  group_map(~calc_self_sim(.x, lyric))

# filter to only the first four tracts
# converting to a factor 
self_sim_96 %>%
  ungroup() %>% 
  filter(track_n %in% 1:4) %>% 
  mutate(track_n = as.factor(track_n)) %>% 
  ggplot(aes(x = x_id, y = y_id, fill = identical)) +
  geom_tile() +
  scale_fill_manual(values = c("white", "black")) +
  theme_minimal() +
  theme(legend.position = "none",
        axis.text = element_blank()) +
  scale_y_continuous(trans = "reverse") +
  facet_wrap(~track_title, ncol = 2, scales = "free") +
  labs(x = "", y = "", title = "Since '96", subtitle = "Lyric self-similarity")

Note that this code was broken as of dplyr 0.8.1