Predicting the next hit track!

Electronic dance music has been on the rise for the past decade and is becoming a trend amongst the youth. More and more neophytes are inspired to produce dance music and its production ahs become easier than ever, due to the availability of free DAW’s[Digital Audio Workstations] and a number of VST’s[Virtual Studio Technology]. Thus, due to this enormously increased competition and a huge availability of these tracks, predicting a hit track by listening to each of them has become virtually impregnable. The record labels thus have to invest enormous amount of money in AnR to quest for the producers who can produce marketable tracks. In 201, music record labels invested a total of 6.5 billion in search of new talent world wide.

Gaining insight of what actually makes a song popular would make finding the hit factor simpler.  Research has been going on for predicting this factor digitally with the help of software, and thereby predict the next hit track. A recent ongoing research by some graduates have utilized various musical as well as practical parameters like tempo, loudness, energy, etc. to predict whether a newly composed song would be popular or not. This is accomplished  by analysing  them with the corresponding parameters – obtained using various algorithms – of the top 40 most popular tracks each month for  the past 40 years’. The big data sets required for comparison are obtained from LAST.FM using its API. The detailed research paper would be available soon.

In the field of Hit Song Science, the intention is to detect, or rather predict, whether a song has the potential to become a commercial hit song, thus reaching the top of the record charts. Quite a few studies have already been done on this topic, some of which focused on extracting general acoustic and lyrical features from songs and then use standard classifiers to separate hits from non-hits. Another approach was to predict potential hit songs by using data mined from a music social network and the relationships between tracks, artists and albums . These studies produced some promising results, but there are still plentiful other approaches to follow. For instance, one could attempt a combination of the acoustic/lyrical features and social features.
The ability to detect potential hit songs would have an incredible impact on the music industry. There are online services, such as uPlaya, which claim to have already succeeded in Hit Song Science, promising that Hit Song Science™ provides immediate feedback on your song’s potential for commercial success and instant legitimization in the market for high-scoring music. This particular online service states that their rating system provides immediate feedback on the quality of music, its competitive edge in the music industry, and its reception among professionals and music lovers. However, even though uPlaya provides objective feedback on a song’s hit potential by comparing it against hit songs from the past, it does not predict whether a song will become a hit: it returns an evaluation based on certain ‘hit characteristics’.

There is still lots of room for improvement on hit song prediction, from which there is much to gain. First of all, it would greatly support record companies in pinpointing songs and artists that have the most potential of scoring a hit song. Furthermore, it could enhance music recommendation services in determining which songs and artists to recommend.

Click here for government certifications