Once we have enough MFCCs of singers, it is possible to obtain a good codebook (maybe vie LBG algorithm) which is able to determine your voice is similar to whom. Vector quantization is necessary to generate the codebook. Then we take the logarithm and do DCT(Discrete cosine transform), delta MFCC can be figured out by this method. Mel filtering will be made to transmit data from regular frequency domain into Mel-frequency. Basically, for a voice data, we first do framing and then transmit them into frequency domain(FFT). To extract the feather of human voice, the best method will be MFCC which is mel-frequency cepstrum coefficients. Users can clearly know which singers' voice are suitable for themselves and they can choose songs from these singers to practice. We aim to design an Android App which provides singer&song suggestions by comparing user's voice feathers with famous singers. As we all know, each person’s voice is unique, but you can still find the similar voice of singers whose songs may be suitable for you. So how to get the right voice guidance and choosing songs are very important. But the question is they might choose wrong songs which are difficult to perform by their voice. Most People are in favor of singing and they hope able to have beautiful songs to get attention from others. Our project name is “Find your voice” and we want to focus on the voice of user.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |