Note: Your custom schedule will not be saved unless you create a new account or login to an existing account.
| Paper: | MSP-P2.3 |
| Session: | Multimedia Systems and Applications |
| Time: | Friday, May 21, 13:00 - 15:00 |
| Presentation: |
Poster |
| Topic: |
Multimedia Signal Processing: Multimedia Database |
| Title: |
CONTENT-BASED RETRIEVAL OF MP3 SONGS BASED ON QUERY BY SINGING |
| Authors: |
Wen-Nung Lie; National Chung Cheng University | | |
| | Chen-Kang Su; National Chung Cheng University | | |
| Abstract: |
With the growing of multimedia in Internet, content analysis of multimedia plays an important role for humanistic management. In this paper, we investigate the content-based retrieval of MP3 songs based on the interface of query by singing. In our method, the MDCT spectral coefficients were directly used to represent the tonic characteristic of a short-term sound. This spectral profile is used for detailed matching between two audio segments. Perceptual features were also computed from MDCT coefficients for audio classification. Two pre-stages based on SVM and k-means classifications were used to remove incorrect (or noisy) segment candidates and speed up following matching process. On the other hand, the schemes of exponential key-scaling and time-warping techniques were developed to overcome key difference and tempo variation between different singers. Experiments show that the retrieving probability of our design can achieve up to 76 % among the top 5 out of a total of 114 excerpts in the database. |
| |
| Back | |