基于深度学习的大数据音频信号增强技术  

Audio Signal Enhancement Technology of Big Data Based on Deep Learning

作　　者：蒋永红 JIANG Yonghong(Guizhou Siso Electronics Co.,Ltd.,Guiyang 550081,China)

机构地区：[1]贵州思索电子有限公司,贵州贵阳550081

出　　处：《电声技术》2025年第2期46-48,共3页Audio Engineering

摘　　要：重点研究基于深度学习的大数据音频信号增强技术,以解决复杂噪声环境下的音频质量问题。该技术融合短时傅里叶变换、多尺度卷积神经网络、长短期记忆网络,构建了一个端到端的信号处理框架。利用大规模多语言音频数据集进行训练,在各种噪声类型和信噪比条件下进行了全面评估。实验结果表明,该技术在实际应用中具有较好的应用效果,为音频信号增强提供了新的技术方案,具有广阔的应用前景。This paper focuses on the audio signal enhancement technology of big data based on deep learning to solve the audio quality problem in complex noise environment.This technology combines short-time Fourier transform,multi-scale convolutional neural network and long-term and short-term memory network to construct an end-to-end signal processing framework.A large-scale multilingual audio data set is used for training,and a comprehensive evaluation is made under various noise types and signal-to-noise ratios.The experimental results show that this technology has good application effect in practical application,provides a new technical scheme for audio signal enhancement,and has broad application prospects.

关 键 词：深度学习 音频信号增强 卷积神经网络 长短期记忆网络 大数据 

分 类 号：TN912.35[电子电信—通信与信息系统]