基于还音转换规则的胶片音频生成方法

Reproduction Transformation Rule-Based Sound Generation for Film Soundtrack

还音作为一种将电影音轨胶片记录的声波信号图转换为音频的过程,是老电影重映中的重要环节.为解决当前还音方法中音频生成失真的问题,提出了一种基于还音转换规则的胶片音频生成方法,可实现音轨胶片图像到音频信号的高精度自动化生成.该方法包含2部分:一是在基于光电转换规则的包络线修正中,基于光度积分的包络线提取,并基于光学规则进行局部与全局校正,解决了光电转换过程中的信号误差问题;二是在基于频率调制规则的音频生成中,设计直流偏量调节方法,并基于电影音频录制还音标准进行信号频域调制,提升了合成音频质量.通过对MovieAD数据集中6类不同类别胶片进行音频生成并测试主观及客观音频评价,结果显示,相比目前较新的还音方法,文中方法生成的音频质量平均提高8.00%以上,且音频采样率可提高到行业标准的162.08%.

Movie reproduction,the process of converting the image-form film soundtrack to wave-form movie sound,is a key step in old movie rerun.Traditional movie reproduction methods are time-consuming because of the cumbersome manual process.The current automatic sound generation algorithms for film soundtrack can ef-fectively save manpower.However,these methods ignore the signal changes in photoelectric conversion process,which result in distortion of the generated audio.To solve the problems,we propose a reproduction transformation rule-based sound generation algorithm for film soundtrack,which enables high-precision audio generation.Specifically,to solve the signal error problem in photoelectric conversion process,we propose an optical rule-based film soundtrack envelope restoration method.To improve the quality of the synthetic audio,we pro-pose a film sound generation method based on frequency modulation rules.Experimental results show that our method achieves a higher reproduction quality compared to state-of-the-art methods.Additionally,the sound sampling rate can be enhanced to 162.08%.