Inspiration
The idea stems from brainstorming around Octave Group's challenge. From there, we knew we wanted to explore natural processing language from a popular series of videos appearing on Youtube - where a popular influencer's clips would be matched with a song.
What it does
Users are able to query the song of their choice, where it returns a video or picture corresponding to the song's lyrics.
How We built it
The queried song by the user makes use of _ Octave's _ music catalogue. With the given information from Octave's API, lyrics are scraped, where it is fed to a natural processing language to return only nouns for the queried song. With the filtered lyrics, it is queried into the _ Shutterstock's _ API, where video/picture assets are returned and concatenated to generate the music video.
Challenges we ran into
The challenge of completing a software project with limited amount of time is common among all participants. Other unique challenges that we have faced is some team members used experimental version of tool(s) which resulted in unexplainable bugs that cannot be resolved and wasted a lot of time in this time-sensitive project.
Syncing the music with the videos is still a challenge.
Natural Language Processing stemming is not quite accurate.
Accomplishments that I'm proud of
- Successfully rendering a full-length video using Shutterstock's API
- Successfully scraping the lyrics
What I learned
- How to work as a team
- Working with different branches
- Working with different API's
What's next for Music Video Generator
The algorithm will be further refined to generate more appropriate assets for the music video and to properly time the pictures to the lyrics.
Built With
- cheerio
- natural-language-processing
- node.js
- react-howler
- react-player
- react.js
- shutterstock
- touchtunes-api
Log in or sign up for Devpost to join the conversation.