SoundSync advice (oh, hindsight (hearing) is 20/20!!)

Caruso wrote on 12/23/2001, 6:34 AM
Woe is me (whoa is my project) unless someone here can give me some sound (that's sound as in wise, sound as in audio, LOL) suggestions!!

Right or wrong, here is what I've done . . .

Son and daughter presented major portions of a classical music recital performed at Steinway Hall, NYC this weekend . . . asked dear old Dad to make a video/audio recording.

I took three cameras, one tripod mounded with two external microphones. The tripod mounted camera was unattended, no zooming, etc. Sound track is fine (whew! at least I'll be able to make a good CD of the performance).

I figured on using the video footage from the tripod-mounted camera as base footage, augmented "tastefully" from close-up shots from the other two non-tripod mounted cams (I never considered using the sound from the outboard cams in the final production, predicting, correctly, that they would be full of mechanical imperfections . . . sound of the tape transport, operator hands, etc.).

(shoulda-coulda-woulda): If I had it to do over again, I would have run patch cords from the audio outs of that stationary cam to the audio line inputs of the other two cams to insure sound sync (and eliminate mechanical sound glitches), but, in reality, what I did was to assume that I could use the sound track from the stationary cam as a "base" and sync/insert video only from the other two cams as required/desired via VV30.

My approach: I've only captured the very first section of the concert (my daughter singing for approximately 30 minutes). I captured that section from all three camcorders, imported each avi onto the Vegas 3.0 timeline, synced the sound tracks of each by lining up the first sound of the piano as a reference point.

Results: If her part were instrumental (thank goodness my son's part in this program only involves instruments), I don't think I would have a problem with sound sync. But, to my eyes, clips from the outboard cams of my “little girl’s” vocals are not as perfectly in sync with the soundtrack from the stationary cam as I'd like them to be, and, I've expanded the timeline to the max, so, unless someone clues me into something I've not tried already, more precise sync is not possible (unfortunately, the most desirable sound record of this performance comes from that stationary cam, the best video shots come from the outboard cams).


Mind you, sync is not terrible, just not perfect (anyone who annually accepts the Orange Bowl halftime show as “real” would have no problems with my sound sync). Obviously, the real time duration of this program varies a bit from one machine to the other.

I don't expect this project to be perfect, but would like input from this board as to how I might make it better.

IF (WHEN) I could do it again:

I think I would run audio patch cords from that stationary machine to the inputs of each outboard cam so that each would record the same exact sound (the distance between these three cams was not great enough to introduce a lag due to sound's relatively slow travel time).

SUGGESTIONS WELCOME ON:

1) What I can do with my present footage (I'm having a lot of fun with it right now . . . I have six "source" tracks, one each for video/audio from each of the three cameras, and one video track I call "final" where I paste sections cut from those three source audio tracks). Video footage cut from the stationary cam obviously syncs perfectly with that cam's sound track. Video footage from the other two cams comes darned close, but not as close as I would like it . . . . any suggestions on further attempts to refine alignment to make for better sync between video of outboard cams and audio of stationary cams (when I copy from the video timeline of either outboard cam, I use markers to insure that I don't lose whatever sync I already have, ok?)? I expect to use the video from that “final” track, and the sound from the stationary cam to render my final production.

2) the soundness of my after thought on linking the source audio by running patch cords from the audio out of the stationary cam to the audio in of the outboard cams . . . seems like it should work to me.

3) how a pro would approach this situation . . . obviously, I'm just an amateur . . . but, I'd enjoy hearing about how someone who does this stuff for a living would have done it. I feel fortunate that I chose to set up that stationary cam . . . the video is bland, not tight enough, but, the sound is grand (good mics, no mechanical noise (as in thumps caused by an all-thumbs operator fumbling for the "manual" auto focus button on one of the outboard cams)) (my muscles are really sore from hand-holding that @#$ camera for 2 ½ hours, ouch!!!!) . . . please, be specific with your comments about what I did (right or wrong . . .), I'd love to hear.

4) Any suggestion on what I may have missed in the way of VV30 features that will make what I have come out better. I believe I've expanded the timeline as far as I can in order to compare the wave forms from each of the three machines . . . I reached a point where expansion of the timeline was great enough that further precision in positioning the cursor was not possible (it would jump from a silence point to a point well beyond the sound commencement point), so, if more precise alignment of these three disparate soundtracks is possible, I don't know how to access that precision.

The good news is that I basically obtained an excellent audio record of this performance (that's really why I was invited to record) . . . and, of course, this new footage is a great holiday toy for me to play with . . . now, if there is any way to make my video/sound sync more precise, well, that would make my day.

Thanks to anyone who has taken the time to read this diatribe, and, thanks in advance to anyone who cares to make a reply.

But for this excellent and flexible (and affordable) VV30 program, I would not be having this "problem" right now, thank you, SF.

Caruso

Comments

wvg wrote on 12/23/2001, 9:39 AM
Unless I'm misreading what you want to do, there is a simple solution. Begin by making an event of whatever video source you want to use. I use the 'split event' feature under Edit. Once you have your 'event' right click, under Group select 'remove from' which will seperate the audio from the video. Place the source from the 2nd camera on the timeline again make an event, use 'remove from' to seperate the the unwanted audio or video.

You now should have the "best" audio and "best" video sources on the timeline. Zoom in, align, once you're happy they sync as best as they are going to you can make a 'new group' (right click again) and have them as one video/audio source you can move anywhere on the timeline or do whatever you want to do with it. :-)
rwsjr wrote on 12/23/2001, 9:50 AM
Your message made me think about the tape some family members shot of my wedding. My dad set up a camera on a tri-pod in the corner of the church and an uncle had a "roving" camera. Neither dad or uncle was working together--just so happens they both decided to tape the wedding and give us the tape. The tapes are now 10 years old and I'm thinking about pulling the two tapes into one production that we could actually sit down and watch.

I'm not a professional, but I've put some thought into your situation and I would get one independent audio recording device and use that as the main audio sound track and mix in audio from the other camcorders only when it was good enough or appropriate to be used. I would get a long enough microphone cord so it doesn’t pick up any mechanical sounds from the recording device. I wonder if there is some inexpensive device out there that can record decent mp3 sound as a substitute for the camcorder sound track?

I don't own a digital camcorder but the two analog camcorders I have owned have terrible sound quality (one records in hi-fi). The main problem is that I pick up too much sound from the mechanical operation of the camcorder as well as other background sounds.

As far as Vegas is concerned, you can use the split function to split the sound tracks from your various sources and use the best one at any given point in time.

One thing I have learned in this hobby---the pros put at least as much time and attention into the audio aspect of their productions as they do collecting the images, probably more. If you add Foley sound effects and try to find the right musical mix, I think the sound becomes more complicated than the picture.

Anyway, good luck!
FadeToBlack wrote on 12/23/2001, 4:34 PM
FadeToBlack wrote on 12/23/2001, 5:37 PM
haywire wrote on 12/23/2001, 6:26 PM
Seems to me you just need to sync video events, not audio events. The simple fact that each camera was a different distance from the audio source is the reason for all your video events not lining up with your audio events. Sound doesn't travel quite as fast as light does. Use the audio from your tripod mounted camera and put up the video only from your handheld shots. Then fade them in when you think it's tasteful to do so, and nudge them until you're happy with the sync.

Good luck
Michael
Caruso wrote on 12/23/2001, 10:18 PM
Thanks for all the replies. Actually, I made a second try at synchronizing the video, and everything worked out nicely.

I shot this performance last night, and I did run all three cams non-stop throughout the entire performance, stopping them only at intermission, at which time I changed tapes (well, actually, my stationary cam records video in analog (2-hrs on the standard 8mm tape) although it has a fine digital recording module and has two RCA mono mike inputs (right & left channels). These mikes were well separated and away from any mechanical noise produced by the cameras (NYC busses and trucks are another matter, however).

It was actually less difficult than I originally thought to pick a well defined moment in the music on the master track, drop a marker, locate the same moment in the audio of the other two cams, drop separate markers (temporarly) at those spots so that I could size each sound track to its marker (using the snap to marker function), then, drag the audio with its audio to the "master track" marker, and, voila, near perfect alignment/sync. I showed the finished product to my family (including my daughter who had performed), and no one could detect a sync problem.

For what its worth, I placed my "master" video/audio on tracks 1 and 2, then the other two cams on 3/4 and 5/6 respectively. I then inserted a blank video track above these six tracks.

This allowed me to alternately mute any two of the three video tracks to view the remaining one (I ungrouped all). Whenever I selected video sections to be included in the final production, I would drop markers around that section, then copy/paste it to the "blank" video track, using the markers to snap the pasted section so as to maintain sync between video and audio.

Of course, once you have your video section properly placed on the "blank" video track, it's a simple matter to resize it one way or the other to create crossfades between clips.

When I had finished pasting selected clips to that top "blank" video track, I simply muted each of the three original video tracks along with the two sound tracks from the non-staitionary cams, and the result was an unbroken audio track running beneath perfectly synced clips selectively taken from all three cameras.

Absolutely worked like a charm . . . and this is the first time I've ever tried this three camera approach . . . Vegas does make it quite simple.

Refreshingly, Vegas never crashed one time during an entire days work, and the final production rendered and printed from the timeline perfectly the first time . . . not one glitch in audio or video.

I do love VV30, thanks, SF.

Sorry to ramble on so, all this is probably old hat to most of you, but it brand new and great fun for me . . . and I impressed the @#%% out of my family.

Caruso --> Additional observations:

Given that my drives weren't empty when I started, storage space did become a consideration. Without thinking, I directed two capture sessions to one of my FAT32 drives, and my final rendering session to the same FAT32 drive (you'd think I'd learn, duh). The two capture sessions were not seamless. They occured in the middle of long contiguous musical passages, and, except that I could make do with footage from the third NTFS capture session at the break points, would have had to recapture each FAT32 session to an NTSF drive.

You'd think I'd learn my lessong (oops . . . lesson) from the capture sessions, but, again, I absent mindedly assigned the final project output render to open space on a FAT 32 drive. Surprisingly, VV30 rendered my 30-minute production to two files which, when played from the timeline were absolutely seamless. I printed output to my DV cam from that timeline, and the file break point is not detectable . . . how 'bout that VV30!! Love it.

C --> sorry to run on so . . .

but I just captured some new footage to FAT 32 exceeding the limit. This capture session (VV30 split into three separate clip files) is absolutely seamless. Don't know what happened the first two times around, but I'm all smiles now.

And, I'm not posting further for now . . . C
Caruso wrote on 12/25/2001, 6:44 AM
I just finished editing the second half of the performance I taped Friday evening. My experience gained from putting that three-camera shoot together using VV30 makes me confident that I can successfully sync video from three sources to anyone of the three sound sources on a regular basis.

If you're interested, here's what I did:

Obviously, one plans on running all cams nonstop throughout the performance (not absolutely necessary, non-identical, contiguous source tapes just means that you'll need to repeat the syncing process more often).

I ran my three cams continuously.

Then, of course, I chose a clear reference point common to all three recordings - someone mentioned a flash, I'm a musician and this was a musical performance, so I used specific features of the audio as reference points.

I isolated a common point in each recording by a process of "hit or miss" by cutting away a slice near where my reference tone should start in the first recording (this was the stationary cam with external mikes which, theoretically, should give me the best audio rendition of the performance). I cut a slice from the wave form, pasted it onto a separate track, and listen to it to see if my reference tone was present. I did this one cursor movement (frame??) at a time until the tone finally showed up in one of my cut/pasted clips. When the tone made itself known, I "undid" that final paste, cut, split to restore the clip containing the earliest incidence of my reference tone, set a reference marker at the beginning of that clip, repeated the same process for each of the remaining two recordings, the only difference being that, once having isolated the earliest incidence of my reference tone in the other two recordings, I snapped them not to a new marker, but to the reference marker set for the first recording.

The challenge for me, when searching for a common sync point between three clips recorded on separate camcorders, is to assure myself of precision. The vv30 timeline only expands so far, at which point, the cursor has a "mind of its own", moving in predetermined increments (I presume the finest precision to be a single "frame" of video (whatever that really is)). At max timeline expansion, my cursor, if moved to screen right, would jump beyond what appeared to me to be the beginning of the desired reference waveform. Move the cursor to the left, and it would jump beyond where the waveform as shown on my screen should begin.

Isolating one frame at a time (by cutting and pasting to a new track) allowed me to prove aurally to myself where the first incidence of my reference sound actually occurred. The precision capability of my present equipment (I do believe) will not allow me to be more precise than that. Of course, I kept the video/audio for each tape grouped so as to maintain sync integrity within each recording.

If your eyes are better than your ears (definitely not true in my case), you could use a flash or some other visual cue, the principle is the same.

Sorry to get so excited and worked up over this issue, but, solving it has made me an instant hero in this here household, so, I figured I'd share it with other users.

As a consequence of my early results using VV30, I already have several requests to produce Tchaikovsky preliminary audition tapes (this would be for pay). Not sure I want to bother with these requests (very sure I'm not ready to give up the day job), but, I am most certainly having great fun with VV30.

For the first time in my (albeit amateur) video editing career, I can actually take raw footage, visualize the end result, and use a computer editing tool to realize that visualization from start to finish . . . no aborted efforts, no finished projects that won't render, no glitches in print-to-tape, no dead-ends . . . just smooth (if time consuming) progress from start to finish.

Oops, looks like I rambled again, but, happy holidays to all, and, thanks SF for a great software package.

Caruso
osiase wrote on 12/27/2001, 3:47 PM
You had asked what the pros do for sync'ing video and audio; they generate SMPTE timecode from one machine that is fed to all video and audio devices (nowadays to the slate as well). For most amateurs (like myself) I can't afford that type of equipment. I had a similar experience taping the Nutcracker then overdubbing a CD (not the same CD used in the performance). After much creative editing, I produced a video that was very good. Only the dancers in the performance knew the music was off.