Implementation of a practical query-by-singing/humming (QbSH) system and its commercial applications

This paper proposes a practical query-by-singing/humming (QbSH) system that retrieves polyphonic music such as an MP3 and its commercial applications. The performance of music retrieval system is mainly affected by the server. This paper discusses developing the state-of-the-art server side software stack which has several managers and plug-in engines. It also describes implementation of digital signal processor (DSP) module for stand-alone embedded platforms. The paper shows three different models for its commercial applications like smart phone, laptop and karaoke. We evaluate the performance of the proposed system with polyphonic music datasets using users' humming datasets as the input query.