2021-10-31
After quite some time I have decided to no longer maintain or support the genius
package. While this package serves a very important purpose from the perspective of music information retrieval, it lies in a grey legal area—web scraping.
Over the years genius.com has changed their web practices in such a way that makes it increasingly unreliable and difficult to scrape.
Why not use the API (as many have asked)? Because song lyrics are owned by the musicians themselves and as such, they cannot be provided via their API.
I will be removing this package from CRAN.
This package was created to provide an easy method to access lyrics as text data using Genius.
Access to genius as an API or via Python here
This package is available from CRAN.
install.packages("genius")
Load the package:
library(genius)
library(tidyverse)
Get started with a tutorial!
learnr::run_tutorial("genius_tutorial", "genius")
genius_album()
allows you to download the lyrics for an entire album in a tidy
format. There are two arguments artists
and album
. Supply the quoted name of artist and the album (if it gives you issues check that you have the album name and artists as specified on Genius).
This returns a tidy data frame with three columns:
title
: track nametrack_n
: track numbertext
: lyrics
emotions_math <- genius_album(artist = "Margaret Glaspy", album = "Emotions and Math")
## Joining, by = c("track_title", "track_n", "track_url")
emotions_math
## # A tibble: 370 x 4
## track_title track_n line lyric
## <chr> <int> <int> <chr>
## 1 Emotions And Math 1 1 Oh when I got you by my side
## 2 Emotions And Math 1 2 Everything's alright
## 3 Emotions And Math 1 3 Its just when your gone
## 4 Emotions And Math 1 4 I start to snooze the alarm
## 5 Emotions And Math 1 5 Cause I stay up until 4 in the morning
## 6 Emotions And Math 1 6 Counting all the days 'til you're back
## 7 Emotions And Math 1 7 Shivering in an ice cold bath
## 8 Emotions And Math 1 8 Of emotions and math
## 9 Emotions And Math 1 9 Oh it's a shame
## 10 Emotions And Math 1 10 And I'm to blame
## # … with 360 more rows
If you wish to download multiple albums from multiple artists, try and keep it tidy and avoid binding rows if you can. We can achieve this in a tidy workflow by creating a tibble with two columns: artist
and album
where each row is an artist and their album. We can then iterate over those columns with add_genius()
.
Pipe a dataframe with a column for the album artists and album/track information. The argument type
is used to indicate if the dataframe contains songs or albums
# Example with 2 different artists and albums
artist_albums <- tribble(
~artist, ~album,
"J. Cole", "KOD",
"Sampha", "Process"
)
artist_albums %>%
add_genius(artist, album)
## Joining, by = c("track_title", "track_n", "track_url")
## Joining, by = c("track_title", "track_n", "track_url")
## Joining, by = c("artist", "album")
## # A tibble: 1,319 x 6
## artist album track_title track_n line lyric
## <chr> <chr> <chr> <int> <int> <chr>
## 1 J. Cole KOD Intro (KOD) 1 1 Can someone please turn off my …
## 2 J. Cole KOD Intro (KOD) 1 2 My thoughts are racing all the …
## 3 J. Cole KOD Intro (KOD) 1 3 There is no reason or no rhyme
## 4 J. Cole KOD Intro (KOD) 1 4 I'm trapped inside myself
## 5 J. Cole KOD Intro (KOD) 1 5 A newborn baby has two primary …
## 6 J. Cole KOD Intro (KOD) 1 6 "Laughter, which says, \"I love…
## 7 J. Cole KOD Intro (KOD) 1 7 "Or crying, which says, \"This …
## 8 J. Cole KOD Intro (KOD) 1 8 There are many ways to deal wit…
## 9 J. Cole KOD Intro (KOD) 1 9 Choose wisely
## 10 J. Cole KOD Intro (KOD) 1 10 At the bottom of the hourglass
## # … with 1,309 more rows
This can be easily replicated with multiple songs as well.
# Example with 2 different artists and songs
artist_songs <- tribble(
~artist, ~track,
"J. Cole", "Motiv8",
"Andrew Bird", "Anonanimal"
)
artist_songs %>%
add_genius(artist, track, type = "lyrics")
## Joining, by = c("artist", "track")
## # A tibble: 102 x 5
## artist track track_title line lyric
## <chr> <chr> <chr> <int> <chr>
## 1 J. Cole Motiv8 Motiv8 1 You really wanna know who Superman is?
## 2 J. Cole Motiv8 Motiv8 2 Watch this, pow!
## 3 J. Cole Motiv8 Motiv8 3 I like him
## 4 J. Cole Motiv8 Motiv8 4 I think he's pretty cool
## 5 J. Cole Motiv8 Motiv8 5 He's my idol
## 6 J. Cole Motiv8 Motiv8 6 I can't have no sympathy for fuck…
## 7 J. Cole Motiv8 Motiv8 7 All this shit I've seen done made my b…
## 8 J. Cole Motiv8 Motiv8 8 Spill promethazine inside a double cup
## 9 J. Cole Motiv8 Motiv8 9 Double up my cream, now that's a Doubl…
## 10 J. Cole Motiv8 Motiv8 10 Please don't hit my phone if it ain't …
## # … with 92 more rows
If you want only a single song, you can use genius_lyrics()
. Supply an artist and a song title as character strings, and voila.
memory_street <- genius_lyrics(artist = "Margaret Glaspy", song = "Memory Street")
memory_street
## # A tibble: 27 x 3
## track_title line lyric
## <chr> <int> <chr>
## 1 Memory Street 1 Ring the alarm
## 2 Memory Street 2 I'm on memory street
## 3 Memory Street 3 With him on my arm
## 4 Memory Street 4 And my feet on the dash of that car
## 5 Memory Street 5 I don't dare
## 6 Memory Street 6 Walk down memory street
## 7 Memory Street 7 Why remember
## 8 Memory Street 8 All the times I took forever to forget?
## 9 Memory Street 9 Call the guards
## 10 Memory Street 10 I'm at the gates
## # … with 17 more rows
This returns a tibble
with three columns title
, text
, and line
. However, you can specifiy additional arguments to control the amount of information to be returned using the info
argument.
info = "title"
(default): Return the lyrics, line number, and song title.info = "simple"
: Return just the lyrics and line number.info = "artist"
: Return the lyrics, line number, and artist.info = "features"
: Returns the lyrics, line number, artist, verse, and vocalist if available.info = "all"
: Return lyrics, line number, song title, artist.
genius_tracklist()
, given an artist
and an album
will return a barebones tibble
with the track title, track number, and the url to the lyrics.
genius_tracklist(artist = "Basement", album = "Colourmeinkindness")
## # A tibble: 10 x 3
## track_title track_n track_url
## <chr> <int> <chr>
## 1 Whole 1 https://genius.com/Basement-whole-lyrics
## 2 Covet 2 https://genius.com/Basement-covet-lyrics
## 3 Spoiled 3 https://genius.com/Basement-spoiled-lyrics
## 4 Pine 4 https://genius.com/Basement-pine-lyrics
## 5 Bad Apple 5 https://genius.com/Basement-bad-apple-lyrics
## 6 Breathe 6 https://genius.com/Basement-breathe-lyrics
## 7 Control 7 https://genius.com/Basement-control-lyrics
## 8 Black 8 https://genius.com/Basement-black-lyrics
## 9 Comfort 9 https://genius.com/Basement-comfort-lyrics
## 10 Wish 10 https://genius.com/Basement-wish-lyrics
genius_lyrics()
generates a url via gen_song_url()
to Genius which is fed to genius_url()
, the function that does the heavy lifting of actually fetching lyrics.
I have not figured out all of the patterns that are used for generating the Genius.com urls, so errors are bound to happen. If genius_lyrics()
returns an error. Try utilizing genius_tracklist()
and genius_url()
together to get the song lyrics.
For example, say "(No One Knows Me) Like the Piano" by Sampha wasn't working in a standard genius_lyrics()
call.
piano <- genius_lyrics("Sampha", "(No One Knows Me) Like the Piano")
We could grab the tracklist for the album Process which the song is from. We could then isolate the url for (No One Knows Me) Like the Piano and feed that into `genius_url().
# Get the tracklist for
process <- genius_tracklist("Sampha", "Process")
# Filter down to find the individual song
piano_info <- process %>%
filter(track_title == "(No One Knows Me) Like the Piano")
# Filter song using string detection
# process %>%
# filter(stringr::str_detect(title, coll("Like the piano", ignore_case = TRUE)))
piano_url <- piano_info$track_url
Now that we have the url, feed it into genius_url()
.
genius_url(piano_url, info = "simple")
## # A tibble: 12 x 2
## line lyric
## <int> <chr>
## 1 1 No one knows me like the piano in my mother's home
## 2 2 You would show me I had something some people call a soul
## 3 3 And you dropped out the sky, oh you arrived when I was three year…
## 4 4 No one knows me like the piano in my mother's home
## 5 5 You know I left, I flew the nest
## 6 6 And you know I won't be long
## 7 7 And in my chest you know me best
## 8 8 And you know I'll be back home
## 9 9 An angel by her side, all of the times I knew we couldn't cope
## 10 10 They said that it's her time, no tears in sight, I kept the feeli…
## 11 11 And you took hold of me and never, never, never let me go'Cause n…
## 12 12 In my mother's home
This package works almost entirely on pattern detection. The urls from Genius are (mostly) easily reproducible (shout out to Angela Li for pointing this out).
The two functions that generate urls are gen_song_url()
and gen_album_url()
. To see how the functions work, try feeding an artist and song title to gen_song_url()
and an artist and album title to gen_album_url()
.
gen_song_url("Laura Marling", "Soothing")
## [1] "https://genius.com/Laura-Marling-Soothing-lyrics"
gen_album_url("Daniel Caesar", "Freudian")
## [1] "https://genius.com/albums/Daniel-Caesar/Freudian"
genius_lyrics()
calls gen_song_url()
and feeds the output to genius_url()
which preforms the scraping.
Getting lyrics for albums is slightly more involved. It first calls genius_tracklist()
which first calls gen_album_url()
then using the handy package rvest
scrapes the song titles, track numbers, and song lyric urls. Next, the song urls from the output are iterated over and fed to genius_url()
.
As this is my first "package" there will be many issues. Please submit an issue and I will do my best to attend to it.
There are already issues of which I am present (the lack of error handling). If you would like to take those on, please go ahead and make a pull request. Please contact me on Twitter.