Music-Mashup is an application which helps playing a playlist allowing for smooth mid-song transitions while ensuring lyrical and emotional relevance tailored to your personal preferences aka your own “in-house DJ”.
In today's world, on one hand there are applications like Spotify, SoundCloud etc. which provide great playlists for every mood and event but DO NOT allow smooth mid-song transitions and on the other hand there are applications like Pacemaker which allow for cropping and editing of music files but require in depth domain knowledge and are not automated. Music Mashup covers all the limitations present in the existing systems and has the following core features:
-
Smart Queueing: Grouping relevant songs based on tempo, genre tag and emotion analysis.
-
Identifying Highlights: Selecting highlights of a song based on lyrical analysis and repetition to create audio slices.
-
Transitions: Ensuring smooth transitions based on tempo.
-
Flexibility: In case of multiple ordering options allows for refresh option which creates a unique mashup each time.
Music-Mashup takes the following steps to generate the mashup :
- Read audio file and convert to wav format
- Retrieve the lyrics for a song by web crawling
- Extract highlights of songs by lyrical analysis
- Calculate BPM for each frame of audio files and relatively order the audio slices using the avg. BPM to ensure smooth transitions
- Play the ordered audio slices using Pygame
In our implementation we use Web Crawling to extract the required lyrics from the website LyricWikia (http://lyrics.wikia.com/). The crawler takes as input the name of the song and the artist of the song and returns the lyrics of the corresponding song.
Beats per minute (bpm) is used as a measure of tempo in music. Methods to find the BPM include:
- Using tools like Audacity
- Implementing a BPM calculator
- For our implementation we use a BPM detection algorithm as presented in the paper of G. Tzanetakis, G. Essl and P. Cook "Audio Analysis using the Discrete Wavelet Transform"
- This gives us the flexibility to find BPM per frame (since avg. BPM is not always indicative of tempo of song segment) for the selected highlights of song and order them accordingly to ensure smooth transitions
Plot is showing (Total Time Taken - Sum of actual length of songs) aka Total Transition Time v/s the number of songs in the input playlist
| No. of songs | Total Transition Time (in ms) |
|---|---|
| 1 | 0.888109207 |
| 3 | 6.464004517 |
| 6 | 12.29310036 |
| 9 | 17.15087891 |
Music-Mashup is developed by Tapan Bohra and Medha Shrivastava under the supervision of Professor Ling Liu of Georgia Institute of Technology as a part of the course CS6675 (Advanced Internet Computing Systems). For help, please contact tbohra3@gatech.edu or mshrivastava3@gatech.edu