| 講演抄録／キーワード |
| 講演名 |
2015-09-11 09:55
Analysis of Watermarking into CELP Speech Codec Based on Formant Tuning
○Erick Christian Garcia Alvarez・Shengbei Wang・Masashi Unoki（JAIST） |
| 抄録 |
（和） |
Unifying the watermarking method into the code-excited linear prediction (CELP) can enable the CELP to automatically provide protection of the encoded speech. In our previous work, a formant tuning method has been proposed. We have evaluated this watermarking method with respect to inaudibility and robustness. The results showed that the proposed watermarking method could satisfy inaudibility. However, when subjected to CELP, watermarks could not be correctly detected since the encoding and decoding process in CELP has destroyed the embedded watermarks. To increase the robustness of proposed method against CELP, in this work, we directly implement the formant tuning method into CELP. The Log-spectrum distortion (LSD), perceptual evaluation of speech quality(PESQ), and the bit detection rate (BDR) were used to evaluated inaudibility and robustness on the proposed method. The results revealed that the proposed method could satisfy inaudibility and the robustness was increased. These results also suggested that the CELP could automatically protect of the encoded speech with watermarking embedded. |
| （英） |
Unifying the watermarking method into the code-excited linear prediction (CELP) can enable the CELP to automatically provide protection of the encoded speech. In our previous work, a formant tuning method has been proposed. We have evaluated this watermarking method with respect to inaudibility and robustness. The results showed that the proposed watermarking method could satisfy inaudibility. However, when subjected to CELP, watermarks could not be correctly detected since the encoding and decoding process in CELP has destroyed the embedded watermarks. To increase the robustness of proposed method against CELP, in this work, we directly implement the formant tuning method into CELP. The Log-spectrum distortion (LSD), perceptual evaluation of speech quality(PESQ), and the bit detection rate (BDR) were used to evaluated inaudibility and robustness on the proposed method. The results revealed that the proposed method could satisfy inaudibility and the robustness was increased. These results also suggested that the CELP could automatically protect of the encoded speech with watermarking embedded. |
| キーワード |
（和） |
CELP codec / speech watermarking / formant tunning / inaudibility / robustness / / / |
| （英） |
CELP codec / speech watermarking / formant tunning / inaudibility / robustness / / / |
| 文献情報 |
映情学技報 |
| 資料番号 |
|
| 発行日 |
|
| ISSN |
|
| PDFダウンロード |
|
|