Innehållsbaserad uppsnabbning av video

Detta är en Kandidat-uppsats från KTH/Skolan för elektroteknik och datavetenskap (EECS)

Författare: Joakim Lundström; [2022]

Nyckelord: ;

Sammanfattning: This report demonstrates and explores alternative methods for speeding up video with associated audio. Two methods have been investigated, these have been combined and evaluated. The first one is to find quiet segments in the video and drastically accelerate them. The other method is to use a speech-to-text service to measure speech speed and speed up the content based on the measured speed, this method proved not to be suitable with the developed implementation together with the chosen speech-to-text software. Dynamically adjusting the speed of audio and video after analyzing for quiet segments is a suitable alternative to a constant speed factor. This method has in some cases made the content easier to understand compared to acceleration with a fixed factor at the same audio and video duration, thus the goal of the work has been fulfilled, although not fully.

  HÄR KAN DU HÄMTA UPPSATSEN I FULLTEXT. (följ länken till nästa sida)