基于时频域信息提取的数字音频乐音识别仿真

Simulation of Digital Audio Music Recognition Based on Time-Frequency Domain Information Extraction

为了克服当前数字音频处理存在信号杂乱、音符无法准确识别问题,提出基于时频域信息提取的数字音频乐音识别方法。采集音频信号,提取时频域信息,采用Goertzel算法获取整体标准音频谱序列,并将所获序列能量归一化处理。利用滤波器将未识别的数字音频频谱序列与标准音模板进行匹配,实现外界因素干扰的去除。由最佳匹配结果获得相对应的乐音音符,实现对数字音频乐音的高效识别。实验结果表明:相比其它传统方法,研究方法表现出了显著的优越性,具有较高的音符匹配精度以及良好的识别性能。

In order to overcome the problems of current digital audio processing, such as disordered signal and inaccuracy of note recognition, a method for recognizing digital audio notes based on time-frequency information extraction was put forward. Firstly, audio signals were collected, and time-frequency domain information was extracted. And then, the whole standard sound spectrum sequence was obtained by Goertzel algorithm, and the energy of sequence was normalized. The filter was used to match unrecognized digital audio spectrum sequences with the standard tone template to remove the interference of external factors. Finally, corresponding musical notes were obtained by the best matching results. Thus, the efficient recognition for digital audio musical notes was achieved. Experimental results show that the proposed method has more obvious advantages than traditional methods. In addition, this method has higher accuracy of note matching and better recognition performance.