Source waveform recovery under reverberant conditions is quite important for machine diagnostics and speech quality enhancement. This article investigates cepstral dereverberation without using reverberant transfer functuins.Source waveforms are typically modeled here by a pulse-like signal,pulse-trains,and speech signals.The dereverberantion experiments demonstrate that a pulse- like waveform and a periodic pulse train are recovered from the reverberant minimum phase cepstrum without all-pass components.All- pass components do not include important signatures elements for those source signals.However,aperiodic pulse trains and speech waveforms convey their signatures also in their non-minimum phase components.The recovery for those waveforms requires their all- pass components.The authors also describe how a power envelope inverse filtering method shows promise for speech waveform recovery.
引用
電子情報通信学会技術研究報告. EA, 応用音響 93(397), 29-36.(1993)
雑誌名
電子情報通信学会技術研究報告. EA, 応用音響
雑誌名(英)
Technical report of IEICE. EA
巻
93
号
397
ページ
29 - 36
発行年
1993-12-17
書誌レコードID
AN10164817
権利
(C) 1993 Copyright IEICE. All rights reserved.
フォーマット
application/pdf
著者版フラグ
publisher
論文ID（NAID）
110003284270
出版者
一般社団法人電子情報通信学会
THE INSTITUTE OF ELECTRONICS, INFORMATION AND COMMUNICATION ENGINEERS
最小位相成分とパワーエンベロープに着目した残響音場における音源波形回復
EIC_EA93_397_29to36_Hirobayashi
(2.39MB)
TOPページ