Top mumble rappers5/4/2023 ![]() Often words will be stemmed to new words that are not in the English dictionary, but this should not be a problem as both the API as well as the original variants would be stemmed to the same base. If the original lyrics is “ballers”, but the API captures “baller”, one could argue it should not be punished for it, since you would easily be able to tell the difference from listening to the track. This is to allow more tolerance for the voice-to-text API. For this analysis both the original as API-generated lyrics were stemmed. For example the words “ball”, “balling”, “baller”, “ballers” would all be reduced to the words “ball”. Stemming is a common method in NLP used to reduce words to their word stem, base of root form. Converting numbers to words (21 Savage -> twenty-one savage).Converting slang to English (lil -> little, em -> them,…).Removing unwanted additions by like “”.(source: )įor the pre-processing, the following steps have been applied to both the original and the API-generated lyrics: RappersĪI generated picture of fictional rapper Lil Controlla’. Here you can find an overview of all the artists in the dataset and some key figures. This list contains more than 30 tracks, but since some of them overlap with what was already in the sample, it resulted in only 30 new tracks, giving a total of 180 tracks. So I added 30 more tracks from Complex’s “The Best Rap Song, Every Year Since 1979”. Of Course more rappers and tracks per artist would yield better results, but as said before, the gathering of data was a very work-intensive task and luckily the results will show most rappers are pretty consistent over the course of different tracks.Īlong the way I realized rap music might have also become less comprehensible over time, given the increasing amount of slang (e.g. ![]() This resulted in 150 tracks (15 x 5 x 2). Select first 5 song results when searching artist on Spotify.Select top 15 (on popularity, not an indication of quality).As the manual part took a tremendous amount of time, I had to limit my sample. The basic concept to determine the amount of mumble remained the same:įor getting the tracks I wrote a simple bot, but the trimming to select the right verse and the copy-pasting of the corresponding lyrics was tedious manual work. As this was the missing key, I got my hopes up again for completing this project. The combination of extremely poor results and a ton of other project ideas, made me decide to give up on this one.įast forward to early 2020, I saw a post on Reddit where someone created an online tool to extract vocals from songs. I started working on this project about a year ago, but after running some initial tests it seemed the voice-to-text converter (Google API) really struggled with beats, instruments and other sounds in the background. Now of course, as with any software project, things didn’t go as smooth as planned. Compare generated text with actual lyrics.My idea to test for mumble was pretty simple: So who gets to decide which rappers meet the criteria for being a mumble rapper? Since even the Wikipedia entry describes it as “loosely defined”, I decided to open my data toolbox to get a more objective indication on how much your favorite rapper is mumbling. However, not every example is as clear, and some rappers do not want to be associated with the term “mumble” at all. In my opinion, a good example of mumble rap would be Playboi Carti’s MEH, of which the top YouTube comment reads “Sounds good, can’t wait for the English version.”. Urban dictionary defines mumble rap as “A form of music that sounds like a stroke victim trying to deliver a speech it is often incomprehensible and illogical.”. In the recent history of hip hop a new subgenre known as “mumble rap” has emerged. From primary school till now, I acquired knowledge and experience in many fields including data and statistics, but of course the world of rap kept evolving too. Since receiving my first cd “Big Willie Style” by Will Smith in primary school, I have been listening to rap practically every day for multiple hours. Those who know me, know that I spend a lot of time working with data and listening to rap music. Warning: this article quotes rap lyrics that may contain explicit language.
0 Comments
Leave a Reply. |