具備車載背景音濾除機制之哼唱查詢式音樂播放系統

隨著音樂播放器的進步，使車用娛樂得以更為實用，但既有的車用播放器仍須手動選歌，為提升便利與維持行車安全，本研究使用哼唱查詢式音樂檢索法提供更為便捷的選歌方式，並針對車載環境設計一個對車載背景音做兩層過濾的機制，先以快速傅立葉轉換求取頻譜，並與背景音樣本頻譜進行相似度計算，在第一層中過濾移除相似車載背景音頻譜的頻譜資料，接著以倒頻譜求基頻，由第二層排除人聲頻率外的基頻，進而得出較為純淨的半音語句，再接續以Dynamic Time Warping法進行音樂樣本比對，找出具相近旋律者並播放之。經由實驗數據分析，以本研究設計的機制進行車載背景音過濾，並對系統做最佳設定，使其在有車載背景音之狀況下維持不錯的查詢準確率。

關鍵字

哼唱查詢式系統；音高追蹤；頻譜； Dynamic Time Warping

並列摘要

Due to the developments of music players, the vehicle entertainments for music are more useful. However, most of current vehicle music players must depend on manual control or key in information to select object songs. In this work, Query by Humming is used by the music play system in order to provide more convenient way for select songs. The designs of the system are focused in vehicle background noises, and there is a mechanism provides two-layer filters. In layer 1 of the mechanism, its work is main to use FFT to convert audio signal frames to spectrums, and makes similar measure with background noises samples. In the layer 2, it converts FFT spectrums to Cepstrums to find fundamental frequency, and then it filters out some fundamental frequency values which exceed of vocal frequency range. By these two layer’s filter’s process, it can improve the input be more pure for querying. Finally, the system measures the similarity of the query melody string and sample melody strings by Dynamic Time Warping algorithm, and find the most similar one to play it from the music database. 　 According the result of experiment, this research optimizes the QBH (Querying by Humming) system by our setting and the design of background noises filter. It has shown that this system can maintain well accuracy of querying in the vehicle environment.

並列關鍵字

Querying by Humming System ； Pitch Trace ； Spectrums ； Dynamic Time Warping

參考文獻

[1] Assent, I., Wichterich, M., Krieger, R., Kremer, H. and Seidl, T.,“ Anticipatory DTW for Efficient Similarity Search in Time Series Databases,” Proceedings of the VLDB Endowment(PVLDB), Vol. 2, No. 1, pp. 826-837, 2009

[2] Büchler, M., Allegro, S., Launer, S. and Dillier, N.,“ Sound Classification in Hearing Aids Inspired by Auditory Scene Analysis,” EURASIP Journal on Applied Signal Processing, Vol. 2005, No. 18, pp. 2991-3002, 2005

[8] Keogh, E., Wei, L., Xi, X., Vlachos, M., Lee, S. H. and Protopapas, P.,“ Supporting Exact Indexing of Arbitrarily Rotated Shapes and Periodic Time Series under Euclidean and Warping Distance Measures,” The International Journal on Very Large Data Bases(VLDB), Vol. 18, No. 3, 2009

[11] Unal, E., Narayanan, S. S. and Chew, E., “A Statistical Approach to Retrieval under User-dependent Uncertainty in Query-by-Humming　Systems,” Proceedings of the 6th ACM SIGMM International Workshop on Multimedia Information Retrieval(ACM-MIR), pp. 113-118, 2004

[12] Unal, E., Narayanan, S., Chew, E., Georgiou, P. G. and Dahlin, N., “A Dictionary Based Approach for Robust and Syllable-Independent Audio Input Transcription for Query by Humming Systems,” Proceedings of the 1th ACM Workshop on Audio and Music Computing Multimedia(AMCMM), pp. 37-44, 2006

國際替代計量

具備車載背景音濾除機制之哼唱查詢式音樂播放系統

未授權

主題瀏覽