2020 Fiscal Year Final Research Report
Large scale audio fingerprint searching in the internet using FPGA
| Project/Area Number |
17K00074
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Multi-year Fund |
|---|
| Section | 一般 |
|---|
| Research Field |
Computer system
|
|---|
| Research Institution | Japan Advanced Institute of Science and Technology |
|---|
Principal Investigator |
Inoguchi Yasushi 北陸先端科学技術大学院大学, 情報社会基盤研究センター, 教授 (90293406)
|
|---|
| Project Period (FY) |
2017-04-01 – 2021-03-31
|
| Keywords | FPGA / 音楽電子指紋 / MIDI |
|---|
| Outline of Final Research Achievements |
In this research we develop a method to calculate audio finger print from an audio file sending via the internet and detect the song and controlling it's copyright. As the first result of this research, we establish a method to identify the calculated fingerprint from the billions number of songs in the audio fingerprint database using multi-GPGPUs. The second major result is a new audio fingerprint calculation algorithm. Conventional method calculates the fingerprint using 128k samples form the top of the audio file, thus the algorithm uses only 2.97 second of the song and it decrease the detection accuracy. We proposed a new sampling method that takes fingerprint samples from the whole of the music file, and we improve the detection accuracy. The final major result is to find the possibility of the design using high-level synthesis.
|
| Free Research Field |
並列処理，
|
| Academic Significance and Societal Importance of the Research Achievements |
近年，Winny や WinMXなど，インターネットでデジタル著作物を交換するソフトウェア(ファイル交換ソフト)が広く用いられ，デジタル著作物の著作権侵害が大きな社会問題となっている．ファイル交換ソフトを利用する理由は，必ずしも著作権を侵害することが目的ではなく，音楽等の便利な入手ツールとして利用している要素が少なくない．もし音楽の電子指紋が高速に検出・検索できれば，誰が誰にどの楽曲を送ったか把握でき，許諾の付与や課金が容易に可能になる．このような手軽で合理的な音楽共有のための手段を構築し，社会の豊かさに貢献することが，本研究の最終目的である．
|
