 |
| Pro Tools 9 running on Windows (Photo credit: Wikipedia) |
I'm working on Libertaria: The Virtual Opera right now with Lucinda McNary, Kera Hildebrandt and Patricia Johnston. One of the challenges we have had is syncing audio to singing vocals, especially since a few of our singers have a lot of vibrato.
MS Lip Sync is not too bad, but the program sometimes misinterprets vibrato, reverb, and sustained tones as a single word, leaving you with a puppet sitting there with his mouth gaping open while several words fly by.
One workaround involved Kera rerecording her own version of the audio over the singer's version and then lip syncing to that. It worked great, but was a lot of extra work.
So here are some ways to prep your audio prior to lip sync. You can use a free program like Audacity or an easy program like Garageband for Mac, Cubase, and other audio software. I use Logic, and high end software like Pro Tools and Digital Performer will give you lots of options, too.
What you will be doing is creating an audio file that is specifically for lip sync purposes. You won't use this file in the end audio mix because it is optimized for lip sync. And always remember to type in the lyrics and play with the words to get the sync right.
1) Get a raw audio file. This means single vocal only, no background music or sound effects. Also no audio effects (reverb, echo)
2) Test in MS. If its great, you're golden, but if it needs work, read on.
3) Import into your software.
4) Listen to the audio file a few times and try to distinguish what MS is not understanding. Is there too much reverb or echo? Is the singer slurring words or sustaining tones? Is there a lack of clarity?
5) Try some of the fixes below:
A - Eliminate Reverb and Echo.
B - Apply a Noise Gate (a tool that eliminates silence, hum and quiet noises). This can be helpful when there are extra breathing noises, mic noises, or even words running into each other. You raise the slider until it eliminates all extra noise. You can continue playing around to "force" a silence between slurred words, For example, it can eliminate the tail end of an "s" that goes into the next word.
C - Play with EQ (equalization). Articulation is on the higher spectrum ("S" sounds). You can raise the volume of the higher frequencies by small increments to try to increase the number of clear consonants in the audio. Don't go overboard, though, otherwise you end up with tape hiss.
D - More EQ. This time cut off the low end (100 and under). This "low cut" will eliminate a lot of unnecessary hum and some mic "bumps".
E - Cut and Trim. You can manually insert silence between problem words by simply selecting the section and "inserting silence" or trimming the section. Try a straight cut instead of a fade out to work with MS lip sync.
6)Export in small sections. You want to export in smaller sections, not one long section. Because MS lip sync does not have automation, the setting is set for the entire audio. By having shorter sections, you can optimize the lip sync settings to smaller clips, ensuring better accuracy.
7) Do multiple takes with different settings. A setting may work for 80% of the vocal file. You can do a second version (with a second camera) that has a setting that works best with the remaining 20%. Then cut and edit around the best takes in post.
8) Use camera tricks for some of the worst lip sync issues. Focus on the singer's eyes, move the camera around where the mouth is in focus for the strongest portions of lip sync. Focus on something else if lip sync is off. Too many times, folks keep the camera straight on the speaking puppet, even if 30% of what they say doesn't sync up. Don't keep it there. In music video production, there are tons of mistakes by live people, so the editor sits with tons of footage and cuts around the bad parts. The audience won't care if you focus on the singer's eyes while she is singing. They will care if they have to figure out why in the world the singer looks like she is in a badly translated foreign flick. In music videos and music production it is even more obvious because lyrics have a specific rhythm and beat that goes to the music. When it is off by a slight beat, it is obvious.
Well, I hope these tips are helpful. We are still working out kinks in our own production, which has 70 minutes of music, with at least half of that sung.
Best of luck to each of you!Musically yours,
Sabrina Pena Young
Composer
http://sabrinapenayoung.blogspot.com/
Libertaria: The Virtual Opera
http://virtualopera.wordpress.com/2013/02/...r-sci-fi-opera/
Comments