More random questions from some jerk

[jpastuch]

OK, at the risk of becoming very annoying on these boards, I'd like some tips on boosting low audio on dialogue in post. Let me say some things first:

-I know, GIGO- I'm not asking for "better audio" or a miracle, the mic I used is not very amazing anyway, and at best gets a very flat sound. I just need to boost the audio with the least amount of distortion.

-I cannot reshoot the scene- no time, and it's important the way I captured the scene and interaction with the characters at the time. Maybe some time in the future, but for now, I just want to salvage what I have.

-It was not a good shot- I had to somehow do the audio and camera by myself so I knew it was going to be awful. I don't have any distortion, it's just low dialogue. Between the past two projects, reading things on the board, etc, I've learned a lot more about audio. But this is before, months ago, so now I have to face the music.

That said...


I'm editing this scene in FCP. This scene has dialogue that ends up VERY low, somewhere in the -42 range on the audio meters in FCP. I'm assuming they should be ending up in between -12 and -6db range, but I could be mistaken that this HAS to be followed. The dialogue is the only sound in this scene save for the sound of them playing cards, which was recorded on location.

I've tried raising the db and doubling tracks, even messing around with equalizers and filters to emphasize certain frequncies so it reaches -12db. It's pretty ugly, but it doesn't have a lot of hiss or distortion, just obviously sounds flat. I can sort of live with it, but I'd like to make it a little smoother.

But rather than just raise db levels, my question is, should I be using an equalizer to raise the levels on certain frequencies instead? Again, I'm not looking to make my audio BETTER...but wouldn't I be better off trying to raise some of the mid frequencies a lot more to avoid to low rumble of the bass and the tinniness and hiss of the higher frequencies?

Sorry for being a moron, the project (for a class) is due Monday. I'm going to try and redo the audio or maybe even the scene in the future. so, basically, I just want to know if I'm making sense- keep the original track, boost it somewhat, then overlay the same track again with only mid frequencies and some high frequencies boosted, rather than just boosting everything.

 

[SMuir]

Try to avoid doubling the tracks to increase gain. I've know many film school instructors who teach that technique to their students, but its just a bad idea. Needlessly duplicating dialogue tracks opens you up to a number of frustrating technical problems such as phase cancellation. Instead, use the software faders or effects plugins to bring the gain up to a useable level. I don't know off the top of my head what the effect is called in FCP, but anything involving "gain" or "trim" is a good candidate. "Normalize" is another common function, although this will typically boost the entire clip until the highest peak hits a desired level (eg. -0.1 dB).

While its not exactly critical that your dialogue sit around -12 and -6 dB, staying in that pocket will bring you dialogue levels roughly in line with low-level broadcast television. If you simply mix to a lower level and leave it up to the viewer to crank up their playback volume, you'll quickly run into two big issues: increasing playback level in this way will introduce a great deal more noise into the signal, and as soon as they switch to another channel they'll be blown out of their seats.

Although boosting the mids is basically what you want to achieve, most low/mid/high EQ strips are a bit too impercise for this kind of job. Instead, break it down into three specific EQ tasks:

1) Use a high-pass (aka. low-cut) filter set to between 90 Hz and 120 Hz to roll off all that rumble. In most dialogue there's not a lot of useful stuff down there, and that's where the lion's share of you mechanical noises (wind gusting, handling noise, etc.) will sit. A fairly steep rolloff (-12 dB per octave) is usually appropriate.

2) Take out some of the hiss by using a very gentle (-6 dB per octave) low-pass (high-cut) filter to take out the high frequencies. This'll be a tricky one, since the tape and mic hisses will likely occupy the same frequencies that add sparkle and transparency to the dialogue. Sweep the rolloff frequency around a bit to see where you get the best reduction of hiss without damaging the liveliness of the vocals. The frequency may land somewhere between 12 kHz and 18 kHz. Trust your ears on this.

3) Use a parametric EQ to further sculpt the sound one frequency-band at a time. Graphic EQs may be a good deal easier to visualize, but the ability to sweep the frequency and Q (narrowness of the frequency range) settings will help you find the frequency ranges that you want to tweak. Different characters will benefit from slightly different EQ settings, and the same goes for different scenes and mic positions. Here's an excerpt from "The Mixing Engineers Handbook" that has some handy information: http://www.digitalprosound.com/2002/03_mar/tutorials/mixing_excerpt1.htm

Best of luck with the edit/mix.

 

[10s]

My 2 cents:

Without it being in front of me this can be akward to advise on, but here I go. I would put the audio into an audio editor and look at the waveform. I would roll off using a graphic EQ everything below 100 to 120 Hz. I would boost around 140-145 Hz by 3 to 4 dB, maybe more to thicken the vocal using a 10-15 Hz wide Q range. Also, EQ the highs to kill excessive sibilence and any overly emphasized areas, or use a de-esser and go to town. Keep enough high end edge so the audience can easily discern the dialogue (not muffled) without any concentrated listening effort and remember to have it fit seemlessly with all the other shots.

I would drop amplitude significantly (a zero level might be the best option, & ramping levels out and in between each word) of all areas of non-dialogue (audio in between words) and then after this is done I would boost the amplitude signal up to the desired matching dB level.

On a seperate track reintroduce any room tone (ambience) to match the other shots.

That's all I can think of at this moment given my understanding of your request.

 

[jpastuch]

Thanks, I wasn't sure about the doubling up bit either.

They teach a lot of awful things in film school- I had one teacher advise me to tape black for entire MiniDV tape PRIOR to shooting so that if I had a space inbetween takes, my timecode wouldn't get messed up. I couldn't believe she was teaching that and told her that was a terrible idea, but she insisted that was the right way to do it. I feel bad for the classmates that listened.

EDIT: Thanks 10s, I figured that's what I had to do but isolating the dialogue the way you put it will probably make it sound WAAAAAY cleaner. It'll also take a ton of time but I don't mind. Good advice, thanks again.

 

[ullanta]

FCP has a "gain" audio filter. You can use that (or a few of them if necessary) to bring your levls up to a reasonable amplitude. Once you see how it sounds once boosted, you can use any of the apropriate EQ/noise reduction tricks above....

 

[jpastuch]

Hm, I've simply just been going into the files and boosting the db- I should go for the actual "gain filter" then? I'm going to edit this in Soundtrack when (and if) I REALLY have the time, until then I'm going to do the best I can with FCP's equalizers.

 

[ullanta]

"Into the files"? I'm not sure what that means... but if you mean in the viewer, that only gives you 12dB of gain; sounds like you need more... it's best to use the "gain" to raise the levels to the point where you can then make fine (and possibly keyframed with the pen tool) adjustments in the timeline....

I don't know how "smart" FCP's EQs are, but they may be doing more than the simple gain filter when boosting audio - thus introducing more artifacts. Maybe not... but if you don't have need for the added functionality, use the simpler filter...

 

[Alex H.]

They teach a lot of awful things in film school- I had one teacher advise me to tape black for entire MiniDV tape PRIOR to shooting so that if I had a space inbetween takes, my timecode wouldn't get messed up. I couldn't believe she was teaching that and told her that was a terrible idea, but she insisted that was the right way to do it. I feel bad for the classmates that listened.

It's called blacking or striping your tape stock, and it can be helpful.

Most blacked stock is used as master tape for output, but because a lot of miniDV shooters don't know how to keep time code from getting broken in the field, blacked camera stock can help avoid that.

That said, it adds needless hours to your camera. If you have a deck that can set custom TC, that's the place to do it. Otherwise, pay careful attention in the field to keep your TC from breaking.

I've been editing a feature and the DP couldn't keep track of TC to save his life. I have three reels in a row with the same hour mark on them, and one of them starts at hour 14, jumps to hour 13 in the middle, and then back to hour 14 (the same start point, by the way) near the end. And because of very bad tape logging by the offline editor, I don't know what reel has what footage, since I can find 14:10:35:00 on three different tapes, and twice on one of those tapes. This was especially painful when one of the drives the producer brought went down... recapturing footage added days to the edit.

 

[Noiz2]

I don't know if blacking and striping makes much sense for miniDV. It's been awhile so I may be off here but on 3/4" You blacked everything and striped edit tapes. The blacking was so that there was no "blank" tape, because edit decks would freak out and decide it was the end of the tape the minute they hit "blank" tape unless they were recording. DAT had the same problem so mini DV may also have that glitch and blacking was a way around it. I'm pretty sure that TC is going to get rewritten when you record though. On 3/4" we only striped edit tapes. You had to be in a special mode (assembly edit I think) so that you didn't step on the TC. I've never seen that mode on a miniDV camcorder so I don't think you can not overwrite the TC on the tape.

edit: Almost forgot. The other major reason was that you always wanted (probably still do) to run the tape to the end and back to re-tension and get any loose gunk off the tape. If your going to play it to the end anyway might as well black it at the same time. DATs we used to FF to the end and RW back but that is not nearly as good as playing through. You will get fewer drop outs and odd glitches if you do this, but I can't remember that last time I saw it actually being done.

ANYWAY, back OT. I second the no doubling "trick". In long discussions with seasoned editors it came out that the original reason for teaching the "trick" was that FCP had no gain filter and you could only get 12dB of gain on the timeline. Editors came up with the doubling as a work around. It does* work. The* is because it opens you up to a whole can of worms down the line, but if done right it will boost the volume with out adding weird phasing etc. BUT it is not nec. anymore so use the filter instead.
FCP and STP have a pretty decent noise reduction filter, I would try that over filtering. Or I should say with filtering. I would roll off the non dialog parts of the sound lows and highs, then boost it up to where you want the dialog then run it through the NR plugin (filter in FCP terminology) and see what you get. Because you probably mostly have broadband noise it might work really well.

If this is going to a sound person just leave it alone and let them deal with it. They have better tools and know how to use them. If they don't find someone who actually is a post sound person instead of someone calling themselves a post sound person.

 

[jpastuch]

Nah, this is a one man editing job for my college. It's all me- and I'm short on time. That's why I'm going to just do the best I can today, hand it in on Monday, then take it home and spend a lot of time on the sound over the summer.

I understand the blacking can be helpful, but isn't it always a terrible idea to tape over MiniDV tapes more than once? I've had issues with an old scene interlacing randomly into the newly taped over footage- and that's with only ONE tape over. Hence, I've never understood that- fastfowarding to make sure the tape unwinds correctly makes sense to me, but taping over black just seems risky. Then again, I've only worked on small projects with a few tapes...time codes have never really mattered to me, but I imagine it could be a very very big issue on bigger and longer projects. Isn't there a better way to keep the TC on MiniDV?

 

[Alex H.]

Nah, this is a one man editing job for my college. It's all me- and I'm short on time. That's why I'm going to just do the best I can today, hand it in on Monday, then take it home and spend a lot of time on the sound over the summer.

I understand the blacking can be helpful, but isn't it always a terrible idea to tape over MiniDV tapes more than once? I've had issues with an old scene interlacing randomly into the newly taped over footage- and that's with only ONE tape over. Hence, I've never understood that- fastfowarding to make sure the tape unwinds correctly makes sense to me, but taping over black just seems risky. Then again, I've only worked on small projects with a few tapes...time codes have never really mattered to me, but I imagine it could be a very very big issue on bigger and longer projects. Isn't there a better way to keep the TC on MiniDV?

Interlacing of old footage comes from repeated use of a tape, or problems with the camera's heads. One pass for black & code is not going to cause issues in the field.

Black and Code is an old-school fail-safe. The reason it is sometimes recommended for miniDV is, again, because a lot of miniDV shooters who don't know better have difficulty keeping TC from breaking, and shooting in TC REGEN the whole time takes out the trouble spot.

To keep your TC organized in the field:

- Start each new reel in TC PRESET, and set the hour and zero the rest.
- Roll :30 of bars. Yes, it's digital, but bars can still be helpful for troubleshooting. Also, the first :30 of tape is highly susceptible to damage over time because of the little lump where the tape is anchored into the hub, so this protects your footage over time.
- ALWAYS be sure to roll for at least 5 seconds before you call for action, and for at least 5 seconds after "cut." This gives you ample heads and tails for log and capture.
- If you power the camera down, or remove the tape at any time, or you rewind it to review footage before moving on, or you decide to go back to a previous reel, do an end search to find the last frame of recorded material and then set your camera to TC REGEN. Then use the transport control to back up a few frames into the tail of the last take. The camera will now pick up the TC at that point on the tape, rather than the last value in memory.

That will keep your TC stream solid, and if you are not the one editing the footage it will keep your tape logger and/or editor very happy.

Now, if you are working in REC RUN, then every stop and start will generate a TC break. That's fine. As long as you have your 5-second heads and tails, you will still be able to function in post.

I should clarify that none of this has to do with the actual technology of miniDV. It has to do with the fact that miniDV is popular with beginners, students, and amateurs. I have worked with shooters on BetaSP, DBeta, and even CineAlta who could not keep their TC straight. The process is the same to avoid TC breaks in these higher formats.

 

[jpastuch]

I've had issues after taping over only once, but then again I was not using professional tapes, so I can't say too much about that. But I remember it happening at some point and me freaking out and becoming paranoid about even rewinding tapes at one point. I guess she was right then, though I did have much smarter teachers advising against doing this, though they had alternatives to keep time code straight (which I can't remember at the time).

I also generally work on school equipment (because I'm too broke for a Mac) and it's seriously pretty dicey. Tapes get eaten left and right, frames drop during capture because the heads are clogged, audio messes up. You waste so much time re-capturing stuff. It makes me angry and I've lodged many complaints.

Anyway, I appreciate your help again everyone.