计算机工程与应用 ›› 2012, Vol. 48 ›› Issue (36): 155-159.

• 数据库、信号与信息处理 • 上一篇    下一篇

改进相位声码器的音频时长变换算法研究

汪石农,许  钢   

  1. 安徽工程大学 电气工程学院,安徽 芜湖 241000
  • 出版日期:2012-12-21 发布日期:2012-12-21

Research of audio time-scale modification algorithm based on improved phase vocoder

WANG Shinong, XU Gang   

  1. School of Electrical Engineering, Anhui Polytechnic University, Wuhu, Anhui 241000, China
  • Online:2012-12-21 Published:2012-12-21

摘要: 基于传统相位声码器的音频时长变换具有变换范围宽、能适用于复调音等优点,但是会引入回响现象和瞬态模糊效应等人为噪声。在传统相位声码器的基础上采用相位锁技术来保持垂直相位相干,利用幅度和频率变化量来跟踪正弦分量轨迹的变化,对音频信号中瞬态分量进行检测和保护以维持感知特性,从而有效地抑制了各种人为噪声,大幅度提高了输出音频的质量。实验结果表明,算法性能良好,具有广阔的应用前景。

关键词: 相位声码器, 相位相干, 正弦轨迹跟踪, 瞬态分量

Abstract: The traditional phase vocoder is a popular algorithm for audio time scaling, especially for large modification factors and polyphonic signals, but it can introduce “reverberation” and “transient smearing” artifacts. Based on traditional phase vocoder, phase locking technology is used to preserve vertical phase coherence. Sinusoidal portion trajectory is tracked using variation of amplitude and frequency and transient portion of audio is detected and preserved for keeping perception characteristic, thus various artifacts can be suppressed effectively, and the resulting sound is improved significantly. Simulation results show that the algorithm has better performance and an expansive application prospect.

Key words: phase vocoder, phase coherence, sinusoidal trajectory tracking, transient portion