A method to give your customers the good taste and experience of professional movie soundtrack designers

Kurzfassung

Scientists of Saarland University have developed a method, which automatically finds the appropriate song for a picture, a set of pictures, or a video. The technique uses the expertise of the best soundtrack designers by sampling awarded movies and creating training sets consisting of movie snapshots and the associated background music. Partners that are interested in further development and commercialization of the invention are sought under license/commercial agreements & technical assistance.

Details

Like million others, you probably take dozens of pictures from the last vacation, a recent wedding, the sport competition of your children, or grandma’s 75th anniversary. Presenting these pictures later on (in private as well as on the World Wide Web) can be spiced up with a well chosen soundtrack. Having hundreds or even thousands of songs available in your mp3 collection makes the selection process a cumbersome, time-consuming task requiring a good amount of expertise and taste.
This technology offers a fully automated solution to this problem as well as a range of new possibilities for multi-media applications in general. In other words, it can be the way to connect the worlds of music and images from Web 2.0, Social Media Networks, video-, photo-sharing to standalone desktop or mobile applications.
Scientists of Saarland University have developed a method, which automatically finds the appropriate song for a picture, a set of pictures, or a video. The technique uses the expertise of the best soundtrack designers by sampling awarded movies and creating training sets consisting of movie snapshots and the associated background music. When a user submits an image, the method performs the following two steps:
- It scans the training set and chooses those snapshots (e.g. three different scenes) that reflect the inputed image best.
- Subsequently, it chooses those songs from the user's collection that fit the movie soundtrack at the scenes selected in step 1 best. Optionally, the approach is able to select a list of suitable songs (e.g. ten songs) and leave the final decision to the user. Video scenes and sets of images can be handled similarly by executing the above steps separately for every single image and combining the results based on group-recommendation techniques.
Partners that are interested in further development and commercialization of the invention are sought. Various technology or patent licensing models are conceivable.

Innovative Aspects:
APPLICATIONS:
- Integration into video-sharing portals (e.g. YouTube, Dailymotion, MySpace) or into photo-sharing portals (e.g. Flickr, Picasa)
- Integration into multi-media orchestration tools such as Animoto
- Extension for presentation and image organizing programs (e.g. Microsoft PowerPoint, Google Docs, Prezi, Apple Keynote, iPhoto, Adobe Photoshop Album, Microsoft Picture Manager)
- Standalone application (Desktop and Smartphone)