星期五, 6月 12, 2009

美國專利 7,529,464

這篇四年多前與同事合寫的專利,終於在今年通過審核。

公告號US 2006/0078287 A1
專利名稱METHOD FOR IMPLEMENTING AN ADAPTIVE MIXING ENERGY RATIO IN A MUSIC-SELECTED VIDEO EDITING ENVIRONMENT
公告/公開日Oct 13, 2004
證書號7,529,464
發明人Yiou-Wen CHENG,
Chen-Hsiu HUANG,
Ming-Jun CHEN
專利權人CYBERLINK CORP.
摘要
A method for implementing an adaptive mixing energy ratio between at least two soundtracks of video footage in a video image-editing environment is disclosed, the process being transparent to a user beyond a single step of initiation. The method includes applying at least one video/audio analysis technique to a session of video footage, demarcating the footage into a plurality of segments, determining a mixing energy ratio for each of the segments according to the analysis, and interpolating the mixing energy ratios for the segments to produce a mixing energy ratio profile before applying the mixing energy ratio profile to the session of video footage.

相關文章:中華民國專利 I307606

沒有留言:

lec-1 (2022-05-12) Accelerating deep learning computation & strategies

雖然用 DNN train/predict model 也好一陣子了,但這週才是第一次搞懂 cuDNN 是作什麼的 以前好奇過 tensorflow/pytorch 是怎麼做 convolution 的,FFT 不是比較好嗎? 下面的 reference 就給了很好的解釋: Wh...