基于人工智能的中文语音文本智能编辑系统设计  

Design of Chinese Text and Audio Editing System Based on AI

作　　者：牛嵩峰 唐炜[1] Niu Songfeng;Tang Wei(Hunan Radio ＆ Television Group, Hunan 410000, China)

机构地区：[1]湖南广播电视台,湖南410000

出　　处：《广播与电视技术》2018年第4期56-61,共6页Radio & TV Broadcast Engineering

摘　　要：广播电台和电视台对新闻采访所获得的音频内容有着巨大的编辑工作量,利用人工智能维特比优化算法、时间戳、MFCC声纹识别、语义识别、语音合成和同音替换技术,设计开发出一种基于AI的中文语音文本同步剪辑系统,能够大幅度提升编辑语言类音频内容效率,提高音频新闻发布的时效性。Radio and television stations stand a huge workload of audio edition after news interviews. A Chinese text and audio editing system based on A1 is designed and applied, through Viterbi optimization algorithm, time stamp, MFCC voiceprint recognition, semantic recognition, text- to-speech technology and homophonic substitution. It improves the editing efficiency of audio content and the timeliness of audio news release.

关 键 词：时间戳 同步 人工智能 维特比算法 MFCC声纹识别 语义识别 语音合成 同音替换 

分 类 号：TN93[电子电信—信号与信息处理]