本文主要介紹一套針對HTTP即時串流（HTTP Live Streaming；HLS）與HTTP動態自適性串流（Dynamic Adaptive Streaming over HTTP；DASH）的雲端影音轉碼系統。為了合乎HLS與DASH的標準，每部影片都會切割成較小的影音片段（audio/video segment），而且每個影音片段都會支援數種編碼格式。為了加速影音轉碼的速度，在本系統中，每個輸入的影音檔案會先根據使用者的設定來分割成數個較小的影音片段，接著透過Hadoop將所有的影音片段分散儲存至所有的節點（node）中，並在每個node上執行影音轉碼程式來對儲存在node上的影音片段進行影音轉碼。根據目前實驗的結果顯示，相較於單一PC的情況，在一個擁有3個節點的Hadoop系統上執行影音轉碼，可以節省約37~49%的轉碼時間。

This paper introduces a cloud transcoding system for HTTP Live Streaming (HLS) and Dynamic Adaptive Streaming over HTTP (DASH). To be compatible with HLS and DASH, each film would be fragmented into smaller video segment and each video segment would have multiple encoding formats. To accelerate the processing time of transcoding, the audio/video content would be firstly partitioned into several audio/video segments in this system. After partitioning, each audio/video segment would be transmitted to all nodes (within a Hadoop system) and then transcoded. Compared to the system with single PC, a system with three nodes could decrease about 37~49% of the transcoding time.