6 Replies Latest reply on Nov 23, 2008 7:21 AM by Jim_Simon

    Getting the timing right...

    OsakaWebbie Level 1
      I'm trying to adjust the speed of an external audio recording on the timeline in PPro CS3 so that it lines up with the audio from the camera's mic. The method I was trying to use is to line up something distinctive toward the beginning (visually using the waveforms), then find something else distinctive toward the end, and then play with the Speed/Duration setting and look at it again. But a strange thing is happening to my waveform. After changing the speed to, say, 110%, it might look lined up in the center of the timeline view window, but not to the left or right. Then I scroll left or right and the visual alignment changes! It's as if my screen is 3D and one track is closer to my eyes than the other! I need to trust what the waveform is saying to me if I'm going to get this to work. Is that a bug? Or something to do with the way CS3 generates the waveform? If I change the speed back to 100%, it acts normal again.

      I've never tried doing such a subtle adjustment of audio using Speed/Duration before. I have had this sync problem before with MD recordings, but they were only off by about 10 frames an hour (faster or slower, I don't remember), so I just manually razored it every few minutes (picking silent spots) and moved it by one frame each time. This time I used a new, high-quality digital IC recorder, hoping that its timing would be better... but it's worse! In the first 20 minutes (all I have tried to work with so far, but the whole event is about 70 minutes total, so this will be a big job) it creeps about 2 seconds! So I have to actually use effects in either PPro or Audition 1.0 (the only audio editing software I have - I don't have Soundbooth yet). I suppose the timeline won't lie to me if I do the adjustment in Audition and re-import the WAV file, but it will take forever to do each iteration of opening in Audition, adjusting, saving, and re-importing and verifying in PPro how close to the goal I am.

      Any clues about my waveform behavior, or a better suggestion about how to do this? If I knew the exact ratio by which the two devices were different (an NTSC DV camera and a WAV file from an IC recorder), I could just do it in Audition once, but I only have a rough estimate so far. I really haven't a clue why two digital devices would be different at all, but it is definitely so. I thought perhaps it was the dropframe of NTSC's 29.97 fps, but neither the MD's variance or that of this IC recorder fit the calculation - so much for that theory...
        • 1. Re: Getting the timing right...
          Jim_Simon Level 8
          Don't use the waveform, actually listen. I do this all the time.

          >I really haven't a clue why two digital devices would be different at all

          There is no technological reason for this, so one must conclude that makers of digital audio records simply aren't do their jobs correctly. Their timing is all over the place, unlike DV cams which are pretty much consistent across makes and models.
          • 2. Re: Getting the timing right...
            OsakaWebbie Level 1
            > Don't use the waveform, actually listen.

            I tried that, but I don't think my ear is as accurate as I'd like this alignment to be, especially since there is already lot of echo in the sound sources to begin with (it was a wedding in a big hall). Also, if you hear a bit of double when you listen, how do you know which sound source is which, to know which direction you need to tweek the speed?
            > Their timing is all over the place, unlike DV cams which are pretty much consistent across makes and models.

            Yeah, I noticed that my two totally different cameras are rock-solid accurate with each other. I have done many two camera shoots (although doing it alone makes me a little crazy, running back and forth between the cameras!), and have no trouble with timing until I introduce an additional audio source.
            • 3. Re: Getting the timing right...
              OsakaWebbie Level 1
              An update after working more on this...

              Yikes! I started using more careful techniques with the listen-and-tweek method, and discovered that my previous feeble attempts at it on this clip were way off base. Being more of a visually-oriented person, the fact that the waveforms looked close caused me to believe that what I was hearing was also close (both were of an echoey voice of the pastor preaching, which, with just a little coincidence, could sound similar until you try to make out the actual words). But I was actually listening to completely different parts of the sermon, due to over-compensation. Here's why: I was naturally zoomed way in on the timeline window when I tried my first few percentages, and apparently some sort of bug in PPro causes just the visible part of the waveform to reflect the Speed/Duration change (that was the 3D look I was noticing, and why I was trying numbers way higher than I needed). Since I was seeing only slight movement, I kept increasing my numbers more and more. Now I finally figured out how far off I was and went back to the drawing board.

              I thought, "Hmm, I can't do it visually because of the bug, and I get lost when I try to do it audibly because I don't know what number to start with or how to distinguish the sounds, so let's just do it mathematically." I went back to the original speed, measured the difference precisely, and calculated the exact percentage I think I need (provided that the recorder's speed is constant): 100.17397%. Since I will have to do a lot more editing on the audio tracks later (I have four sources of audio to mix together to try to get all the elements of this elaborate wedding - a Filipino style ceremony done bilingually in English and Japanese!), and since PPro's Speed dialog only handles two decimal places, I thought I'd do the change in Audition and import the properly timed WAV back into PPro.

              Okay, nice thought, Karen, but... I tried both resampling (let the pitch go up ever so slightly) and time shifting (recalculate everything to keep the pitch the same). I heard too many warbling artifacts when I time shifted, but I ran into a bug in Audition when I resampled - although the program says my 59:25-long original became 59:18, as it should, the resulting file after saving is 59:31! And yes, the wave really is stretched - I brought it into PPro and laid it on the timeline, and it's obvious. Argh, this is getting frustrating!

              Edit: I was mistaken - Audition does say the wave is 59:31, and it's because the logic of the number 100.174 reverses when you choose the resample vs. time shift button in the same dialog box! Go figure... So I resampled at 99.826, and it looks like this one will work for me.
              • 4. Re: Getting the timing right...
                Jim_Simon Level 8
                >how do you know which sound source is which, to know which direction you need to tweek the speed?

                In the mixer window, send the nat sound to the left channel, and the mic'd sound to the right channel. Get them lined up at the beginning of the clip as you were doing. Adjust volumes temporarily so they're about equal. You don't want one overpowering the other. Go to the end of the clip, and experiment with adjusting the speed until they also sound in sync at the end of the clip. If the nat side is heard first, the mic'd clip is too slow and needs to be sped up. If the mic'd clip is heard first, then slow it down. Hit CTRL-Z after each test to return the speed to 100% before trying again. When you've got it dialed in, return the balance to center for both tracks, and adjust the volume mix appropriately.

                This is how I do it for weddings.

                I have discovered my recorder's "magic number". It's 100.04% Now I just line them up at the beginning, apply that speed change to the audio and all is good. No one will ever notice any pitch change with such a small adjustment.
                • 5. Re: Getting the timing right...
                  OsakaWebbie Level 1
                  Panning left and right: good idea, thanks.

                  But in the future I think I'll always do the math and process the audio file outside of PPro, because I don't think I can handle having the waveform in the timeline not reflect reality of where in the waveform I actually am. If my recorder is as consistent as yours, I now know its "magic number", so that will make it easier.
                  • 6. Re: Getting the timing right...
                    Jim_Simon Level 8
                    In my experience, the waveform will be accurate once you move around the sequence a little bit, or even resize it.