Audio narration (voice-over) for a video

Easy version

First, I highly recommend that you write out your voiceover *before* ever recording any video. Write down what you want to say, then figure out exactly what you want to show during each section. Time out how long each section takes to say, speaking slightly slower than normal conversational speed. This can be an iterative cycle, until your video and audio ideas match up.

I realize this sounds like a lot of work, but it will make your life and video *much* better if you know how long each video segment should be and make an effort to create them with the correct timing. It's particularly an issue when you're recording interactive sessions off the screen, where if you don't make a real effort to do things according to schedule you'll have boring video that lasts way too long.

If you record your soundtrack before creating your video, then you can play that back to yourself as a timing aid when you're recording the video segments.

More detail

Finding the right mike to plug into a computer

The default "home" for the silver mike-preamp combination is the conference room. The silver mike can usually be found on top of the cabinet, and the box directly connected to the mike is the preamp. The box after the preamp is called a "matchbox" which provides even more amplification when you want to record voice in the conference room onto moviola. But if you just want to record your voice onto any other Indy elsewhere, you don't need the matchbox. So you should find the plug on the matchbox that says "IHF Input" and disconnect the wire from there. This is the output of the preamp that can almost be plugged into the mike input of an Indy. We say "almost" here because the connector at the end of the wire doesn't quite fit the mike input of the Indy. You need to add a converter to it. And for your convinience, we've taped an appropriate converter on the top the preamp box. Just make sure when you're done, tape the converter back to the preamp box. There are a couple of buttons on the preamp and they should all be at the "up" position (rather than "down"). The "gain" knob for "microphone 1" is the only knob you might want to turn in the recording session to give you the disired signel level. According to the tests we did a reasonable start position for the knob is between 40~50dB, if the Indy's audio mike input level slider is left in its default middle position.

Recording your voice

• Type apanel to bring up the audio control panel.
• Select the left side Analog In block by clicking on the left, so that it's rubberbanded in red.
• Pick Selected->Input Source->Microphone.
• Check the Meter and Monitor box.
• Say something into the mike, and play with the audio levels until you can hear yourself in the phones but don't sound distorted. Note that you may have to play with the audio out sliders as well as the audio in sliders to hear yourself. If you're using the silver mike/preamp combination, you can also turn the "gain" knob for "microphone 1" on the preamp box rather than the audio level slider. You want to avoid as much as possible having the input meter go all the way into the red, but occasional yellow is fine.
• Type "soundeditor". Hit the "Record" button and say your voiceover. When you're done, hit stop. You'll then be able to see the waveform of your audio sequence.
• When you click at a point you'll see a vertical line. Hitting Play will play back starting there. You can highlight a section, and then the Play button will play only that piece. (Note that sometime you think you've just clicked, but it thinks you've dragged a bit, and will only play a fraction of a second. Just click again, trying not to move the mouse, to solve this problem.)
• Almost everybody makes mistakes in the voiceover. There are three ways to deal with this:
  • Start all over. Bad idea, and unnecessary.
  • Hit stop, find a good restart point right before the mistake, and start recording again from there. Much better than starting over, but often you'll be holding the mike at a slightly different place and your voice volume will be different, so these breaks can be somewhat noticeable. You can mess with the audio levels of the recorded waveform, but that's nonoptimal.
  • Don't stop recording. Just say the part again that you messed up, and keep going. You can go back afterwards and cut out the bad stuff.
• You can edit the waveform by selecting pieces and using the Edit menu to cut, copy, and paste. If you're going to paste in empty space, be careful to copy/paste a totally empty segment from the very end of the recording. Trying to use a section that looks quiet from the middle of the recording can have bad consequences, since there's usually the noise of your breathing or something there.
• Save your file.

• If you're making the rest of your movie using Adobe Premiere, you can just import the audio file. If you need to manually insert edit it onto the audio track of a videotape, see the documentation for the appropriate deck.

• I recommend that you split your long audio file into scenes (usually you'll have many different video segments, either actual or conceptual, that make up your movie) instead of importing the entire thing into Premiere. It's then easier both to adjust the audio/video synchronization by simply moving those blocks independently, and I think Premiere is much more efficient when making the final movie (i.e. takes one hour to compile instead of two). You can do the splitting in soundeditor - save off the whole file somewhere in case you ever need it, and then select big blocks and copy/paste them into a second running copy of soundeditor. Use "save as", then "clear", then paste the next segement in.
• See the voiceover suggestions on the capture page for details on how to hear what you're saying while you record the live video sections of your video, so that you can get the timing to work out properly.

Additional notes

Old-style recording directly to tape

• One of the many downsides to this approach is that you have to do it in the videolab itself, which adds a lot of background noice to your soundtrack. One of the loudest pieces of equipment that you may with to temporarily power off is the Folsom scan converter. If you turn off the main ventilation system to lessen the noise, PLEASE remember to turn them back when you're done so that our $300K worth of video equipment doesn't melt. (I'm serious - leaving the fans off for an entire day could cause major damage.)
• You need to use the big studio-looking silver mike with the round head, not the PC-oriented black one (they have different output levels). The microphone should be found in the the conference room connected to a "preamp" box. Disconnect the microphone from the preamp input. For the cord with the 3-pin audio XLR jack on the end that connects with the microphone and a "phono" type jack on the end that plugs into the deck, look in the XLR bin of the wooden shelf in the corner of the video lab.
• The audio DUB/CH-1 button on the UMatic Play deck allows you to record your voice into audio channel 1 without disturbing the video track. Plug the microphone into the CH-1 microphone input in the lower left corner of the deck. When you hit the button, you're recording audio. The tape does not preroll, it starts from right where you are. The CH-1 AUDIO LEVEL knob controls the volume level that you record at. I've had good results around 3 or 4, your milage may vary. The needle on the gauge above the knob can go into the red zone on the right occasionally, but should never "peg out" on the extreme right side. Whenever it does so the recorded voice is disorted. It's easy to incrementally record your narration, but it's hard to go back and fix a mistake in the middle of things. I highly recommend that you check after every edit to make sure that everything sounds right before going on. (Note that the audio meter only works when you're actually recording or the deck is at a full stop. It will not work when you're in pause mode.) Also remember that when you're recording the speakers must be turned all the way down or else you'll have feedback problems.

Troubleshooting

• It's possible to run out of disk space for the temporary files created by soundeditor if they're in your already overloaded home directory. You probably want to record them to local disk, though, not moviola's /nle directory. You can explicitly set the temp directory by setenv TMPDIR /usr/tmp, for example. Audio files can be reasonably large (but not as bad as video) - 60MB for 5 minutes, for example.

• Do not try to plug the silver microphone directly into a computer, it won't work. Only the black Labtek mike is "line-level", which means you can plug it directly into a computer - the silver one is "mike-level". We now have a pre-amp to turn the silver one's mike-level output to a line-level signal, which can then be fed into a computer.

  Although it's an extremely bad idea because of the ambient noise in the video room, you could theoretically use the silver mike directly by plugging it into the IO panel and routing that signal to the audio mixer - i.e.
  telect -laudio -iio2 -omxr1-2
  would send the signal to the first two bays on the audio mixer. You would then turn the trim knobs at the very top of the mixer board from the 'U' (unity) setting (9 o'clock on the dial) for line-level to much higher for mike-level - about 3 o'clock on the dial - so that you can see the lights blink when you have the bottom sliders near unity (about 3/4 of the way up). You should read the mixer manual to see how to set the levels properly.

• If for some reason you need to use the black Labtek mike instead of the silver mike-preamp setup, one way to get better sound quality is to cover the mike loosely with a spare shirt or sweater. That will approximate the acoustic baffling built in to the silver mike head, which helps mitigate the popping effect you get from aspirating into the mike when making sound like "p".

Return to Main Video Menu
videolab@graphics.stanford.edu
Last modified: Sun Jan 30 18:16:05 PST 2000