we will be receiving a dozen or more WMV files daily that need random portions of the audio muted. These portions would only be about a minute or less in length. The files will range from an hour to six hours in length. I have two people to do this process. Computer time is free, but the entire computing process must be no longer than the file length, otherwise the jobs won't finish in one day.
Since we will be receiving so many and need to turn them around the same day, I am looking for a process that would not require creating separate clips to edit and re-splice; or demuxing the file, opening the audio stream in another program, doing the muting, remuxing and re-rendering; or any long processes such as those.
Since I can restrict my input to wmv only; does anyone have any suggestions? Would a different input file type make it easier? I pretty much have to deliver wmv files.
p.s. I did a couple searches and did not find any good ideas.
And obviously, if I should post this elsewhere, please let me know.
the points are not really random, they will always be shortly preceded by the words "recess" or "record" (possibly others). But not every "recess" or "record" will not need to be muted. So technically they aren't really random, but there is not any regularity that I could use for any initial programmatic solution. I will, however, have rough time-in-video times to allow a manual muting process to be carried out.
My wish is for a program to allow a tech to load the wmv file (I do not have any need for the load process to create it's ideas of what constitutes a clip), (use the time-in-video to) go to the approximate location, mark the in and out points, mute that audio; go to the next location, etc. Once completed, do a Save As (let the computer do any rendering, etc.), and move on to another file. Something like Audacity is fine to use to mute the audio (we use it for the wmv files), but I cannot incur the time of demuxing and remuxing to process the wmv files in this manner. I also cannot incur the unknown about the audio stream not staying in sync with the video stream.
Please let me know if there is anything I can add to help produce a solution.
Also, since I can't do it to your reply; Cornucopia, I do appreciate your thoughts. Regarding the laws of space and time; if I had a million monkeys typing on keyboards... I could teach them to do this process instead.
+ Reply to Thread
Results 1 to 7 of 7
Last edited by crbeckman; 23rd Jan 2013 at 15:16. Reason: won't let me reply
Format has nothing to do with it.
When you say "random portions", are there any common features to those portions that could be exploited (silence for a certain amount of time, loud burst, tone of certain frequency)? If not, and those portions are TRULY random, you have no other option than these:
1. PLAY the WMV file out to analog, use a mixer IN REALTIME to ride the volume MANUALLY (by operator), while simultaneously doing a live re-encode. Will incure a generational loss in both the A+V streams.
2. Use a network server & multiple workstations & multiple operators and go through each separately in parallel, doing the usual edit/splice/demux/process/mute/re-encode/remux. Each will take a while, but the processes should at least be done in time, in aggregate. Only the audio will incur generational loss, the remux with the video shouldn't affect it (well, depending on what kind of processes - it would help if you were clearer & more detailed).
You can't get blood out of a stone, nor can you ignore the laws of time & space. Sorry,
I would do it using Scott's method #2:
1) import the WMV in a NLE, mute the required audio sections, export the audio track (only) as WMA
2) open the original WMV in Windows Media Stream Editor (comes with Windows Media Encoder) , add the new WMA track, uncheckmark the old WMA track
The reason for the NLE is I'm assuming your assistants will need visual cues ; if not , you might be able to do the 1st step faster in a dedicated audio editor, and/or if they already have the timecodes . Although no time is wasted demuxing, time is "wasted" importing in a NLE (NLE's can be "slow" when then conform audio, building peaks etc...), usually this is "snappier" in an audio editer
Another format that "smart renders" with typical NLE's would be more ideal (e.g. DVCProHD, AVCIntra, XDCAM), because you could do everything in 1 step, 1 program. The video would be passed through untouched (no quality) and multiplexed . But this is not as feasible for you since you are given WMV, and the desired goal is WMV (No NLE smart renders WMV, so you would be incurring losses going back and forth, also it would take much longer)
Yeah, I slow down when I try to post AND do actual work at the same time!
Thanks to both of you (poisondeathray and cornucopia). I'll probably end up grabbing a developer to provide a tool to allow us to easily mark the in and out points (for the muting) and then pass off the rendering to other servers. That keeps the production people using the skills that need a people process and off loads the rendering (non-people process) elsewhere.
I'm lucky in this scenario, as I will already have access to pretty close time-in-video values, an existing program that allows access to the media files and the synced time-in-video values and a developer that can add additional marking points to the program. Also able to not have to process the file(s) except at the end. (Ended up moving from a muting process to a cut-and-render process. Hey, the requirements were able to be changed to fit a better solution.)
I do appreciate your thoughts and efforts. Kudos to both of you.