电脑知识与技术
電腦知識與技術
전뇌지식여기술
COMPUTER KNOWLEDGE AND TECHNOLOGY
2015年
10期
187-189
,共3页
K-L变换%离散余弦变换%自适应增强%数据压缩
K-L變換%離散餘絃變換%自適應增彊%數據壓縮
K-L변환%리산여현변환%자괄응증강%수거압축
K-L transform%DCT%self-adaption enhancement%data compression
为了实现数字语音的有效压缩，提出了一种基于K-L变换的语音自适应压缩算法。由语音帧向量构造协方差矩阵；通过对协方差矩阵的特征值分解，得到与帧对应的正交矩阵；用正交矩阵对帧向量作正交变换得到变换系数向量，并将其截短；由截短的变换系数向量重构帧向量；最后经抽取，实现对帧向量的压缩。在不同信噪比下对不同语音样本进行仿真实验，并同离散余弦变换压缩比较，实验表明，该算法不仅数据压缩率高、重构语音清晰和自然，而且同时实现语音良好的自适应增强。
為瞭實現數字語音的有效壓縮，提齣瞭一種基于K-L變換的語音自適應壓縮算法。由語音幀嚮量構造協方差矩陣；通過對協方差矩陣的特徵值分解，得到與幀對應的正交矩陣；用正交矩陣對幀嚮量作正交變換得到變換繫數嚮量，併將其截短；由截短的變換繫數嚮量重構幀嚮量；最後經抽取，實現對幀嚮量的壓縮。在不同信譟比下對不同語音樣本進行倣真實驗，併同離散餘絃變換壓縮比較，實驗錶明，該算法不僅數據壓縮率高、重構語音清晰和自然，而且同時實現語音良好的自適應增彊。
위료실현수자어음적유효압축，제출료일충기우K-L변환적어음자괄응압축산법。유어음정향량구조협방차구진；통과대협방차구진적특정치분해，득도여정대응적정교구진；용정교구진대정향량작정교변환득도변환계수향량，병장기절단；유절단적변환계수향량중구정향량；최후경추취，실현대정향량적압축。재불동신조비하대불동어음양본진행방진실험，병동리산여현변환압축비교，실험표명，해산법불부수거압축솔고、중구어음청석화자연，이차동시실현어음량호적자괄응증강。
To realize effective compression of digital speech, a kind of speech framing compression algorithm based on K-L trans?form is put forward. Covariance matrix is constructed by speech frame vectors;orthogonal matrix corresponding to frame is got by decomposing eigenvalues of covariance matrix;the transform coefficient vector is got by orthogonal transform of frame vectors with orthogonal matrix, and it is truncated; then frame vectors are reconstructed by truncated transform coefficient vector; finally, the compression of frame vectors is realized by extracting. Under different SNR simulation experiments on different speech, and com?pared with the DCT compression. Simulation experiments show that the algorithm has high data compression rate, reconstituted speech is clear and natural, what’s more, good self-adaption denoising of speech is realized.
