基于遗传算法的智能作曲技术研究

发布时间：2019-05-16 22:42

【摘要】：为了利用遗传算法进行智能作曲,对由计算机自动生成音符序列和音频文件的具体问题进行了讨论。计算机根据预先设定的参数生成初始乐段群体,将对各乐段的人工评估结果作为适应度函数值,分别设定选择、交换和突变规则,通过时值修正来解决进化过程中乐曲每小节各音符的时值之和的不稳定问题,完成了音符序列的计算机生成问题。音频文件的产生通过建模和编码的方法实现。建立表示振幅与频率、时值之间关系的振动模型,其频率因子和时值因子分别取自音符编码中音高分量所映射的频率值和时值分量所映射的时间,根据此模型,结合数字音频技术,最终生成WAV音频文件,从而完成了音符到音频文件的转换。与目前MIDI形态的计算机音乐相比,WAV形态的计算机音乐具有很强的丰富性、灵活性和稳定性。实验表明,利用这种交互式遗传算法所创作的乐曲能基本满足部分人的情感表达需要和审美标准,对促进计算机音乐技术的发展有重要的科学意义。
[Abstract]:In order to use genetic algorithm for intelligent composition, the concrete problems of automatic generation of note sequences and audio files by computer are discussed. The computer generates the initial segment group according to the predetermined parameters, takes the manual evaluation results of each segment as the fitness function value, and sets the rules of selection, exchange and mutation, respectively. The problem of unstable sum of time values of each section of music in the process of evolution is solved by time value correction, and the problem of computer generation of note sequences is completed. The generation of audio files is realized by modeling and coding. A vibration model is established to represent the relationship between amplitude, frequency and time value. The frequency factor and the time value factor are taken from the frequency value mapped by the pitch component and the time mapped by the time value component in the note coding, respectively. According to this model, Combined with digital audio technology, the WAV audio file is finally generated, and the conversion from note to audio file is completed. Compared with the current MIDI form of computer music, WAV form of computer music has a strong richness, flexibility and stability. The experimental results show that the music created by this interactive genetic algorithm can basically meet the needs of emotional expression and aesthetic standards of some people, and it is of great scientific significance to promote the development of computer music technology.
【作者单位】：河南师范大学计算机与信息技术学院;
【基金】：河南师范大学引进博士科研启动基金(No．0716)。
【分类号】：TP18;J614.5

【参考文献】