碩士 / 大同大學 / 資訊工程學系(所) / 95 / nternet has become the most important way to search and obtain the information in the modern world. Not only the data contains the text but also the sounds, image and animation will people wish to look for by using the internet searching engines. Searching a song as an example with our experience before, we usually remember part of the melody in a song, not its name. But what the current searching engines provide is doing by inputting the keywords. In this way, if people can’t remember or input the names of the songs, the singers or the composers correctly, they would not find the ones they really long for. Hence, there comes one of the most popular topics of searching engine with multimedia. Whether it is possible to find those songs or compositions we want by inputting the format of symbol in the searching engine, is also the resolution that many other searching engines would like to upgrade or adopt.
It has been more than one thousand years since music exists with the format of symbol. Stave is the most common example. By reading the staves, singers and performers can easily present the works that composers made it years ago. We may also remember the songs by writing it down with staves. But how can we find the names of the songs by searching with the symbol of the stave? There is something to do with the similarity of melodies. When we input a song, we expect the searching engine to judge and compare by its similarity with those samples in the database, then appear the list of the most similar ones for us to select the one we really want.
This thesis applies the transportation algorithm in the field of Operation Research. It implemented a comparing system which compares the similarity of melodies with songs. By the concept of resolving the transportation problem, we got the cost of transportation. And the cost will determine the levels of similarity. We use the simplex method implementing the algorithm to compare the similarity of what had been input in our database. The output of experiment helps us to see clearly the correlation between the cost of transportation and the similarity of melody.
In this thesis, we briefly explain the theory of transportation problem, and its mathematics of problem resolution. Then we encode the format of MIDI, that is usually recorded as the melody of a song, to be computable. It will help us to find out the most similar song by searching the lowest cost of transportation.
We provide a simple platform for users to enquire by inputting the data with the MIDI format. The tool will list and sort 10 of the most similar songs after computation. In order to speed up the computation, we compare the song input with the group representatives to lower the time complexity. It reduces the response time to the users while they are searching. The platform also provides the music player with staves. That helps the users to listen to the songs listed as well and make sure whether it is what they enquire.
| Identifer | oai:union.ndltd.org:TW/095TTU05392014 |
| Creators | Wei-Yuan Chen, 陳威元 |
| Contributors | Jyun-Ming Chen, 陳俊銘 |
| Source Sets | National Digital Library of Theses and Dissertations in Taiwan |
| Language | zh-TW |
| Detected Language | English |
| Type | 學位論文 ; thesis |
| Format | 36 |
Page generated in 0.0078 seconds