It appears that MP3 meta data does not have an explicit field for duration that is recognized by all media players. It was brought up again by a user here on reddit and that finally made me analyze the matter.
It always depends on many factors, of course, in particular on the chapter structure of the book, but the conversion tests I made with 1.12 were 2 to 3 times faster than before. The program now separates decryption and transcoding very similarly to what it does in “Split chapter” mode. In 1.12 I have modified the internal workflow.FFmpeg, which does the real work here, had more or less to repeat the same stuff for each chapter as if it were the whole book, but actually only needed to convert a single chapter. While (unsplit) “Chapter” mode also uses parallel processing since version 1.3, this has not been very efficient. For MP3, “Split chapter” mode has been the fastest until now.
Performance boost for MP3 output and “Chapter” mode: What’s new in version 1.12? No new features this time, but some changes “under the hood”.
0 kommentar(er)
