全文预览

小数据下的音素级别说话人嵌入的语音合成自适应方法 徐志航

上传者:upcfxx |  格式:pdf  |  页数:15 |  大小:0KB

文档介绍
unstablespeechsynthesizing.Ourproposedphonem elevelР farasw eknow , buildinganend to endTTSsystem w itha speakerem beddingm ethodprovideanewaspectofusingР sm allam ountofdataisstillnotaw ell solvedproblem .Itis differentgrainsspeakerem beddingandim provethequalityofР alsochallengingtobuildanend to endTTSsystem w itha synthesizedspeechesbothonsim ilarityandnaturalness ,Р sm allam ountofdirtyspeech.Adaptationfrom apretrained w hichcanbeappliedoneveryTTSm odelsforspeakerР TTSm odelisapopularandcom m onlyused. adaptation.Р Thispaperw orks onspeakeradaptation m ethodof

收藏

分享

举报
下载此文档