Crowd++: unsupervised speaker count with smartphones
Proceedings of the 2013 ACM international joint conference on Pervasive and ubiquitous computing
Hi-index | 0.00 |
This paper presents a Speaker Count (SC) algorithm based on pitch estimation designed to recognize single-speaker audio recordings from two-speaker recordings. The paper reports the implementation for Symbian OS smartphones. The proposed solution is aimed at limiting implementation complexity and related battery consumption with respect to the schemes available in the literature. This action is mandatory due to the restricted resources available on smartphones. A practical experimental campaign of the proposal is conducted in real smartphones. The Speaker Count method has produced encouraging experimental results. The results also show that the implemented method allows saving energy and acting in a reasonable amount of time.