Basic suggestions for EQ on talking heads of various quality
 

....ok, a little more then EQ might be nice, too.

But I'm working for a hospital that has been recording educational videos of doctors and physicians to be watched by other doctors and physicians. So basically, the audience will most likely be listening through home laptop or crappy desktop speakers.

On top of that, the audio was captured on a lav mic by non-video people often in a noisy air conditioned room.

I occasionally run a de-noise process in Aodbe Audition, which can leave the vocals a bit thin sounding (at best). So I then EQ, and put on a hard limiter at -6db and get the audio to average at -12db. Everything sounds pretty good so far...

... that is, until I realized that my Sony MDR-7506 quality monitor headphones, might not be giving the best representation of how the audio is going to sound on crappy speakers. Low and behold, when I listened through cheap headphones or small cheap speakers, the low end sounded muddy and muffled, masking much of the audio. Bringing that down, the high end sound sharp and piercing. I'm starting to think, screw it, I'm better of not equalizing at all.

Any suggestions or encouragement?

Re: Basic suggestions for EQ on talking heads of various quality
 

The 7506 have a bump in the mid-range that is great for field recording; you can really hear what's going on with vocal recording.

But, for the reason you've discovered, they are not a good choice for mixing.

In fact, no headphone is good for reference in mixing, unless you're mixing for headphone playback.

Reference-level monitor speakers are what you need. Discussed many times on this forum, search is your friend. I favor small Genelec boxes for dialog, but there are many other good choices, not all of them spendy.

Re: Basic suggestions for EQ on talking heads of various quality
 

I'd say your medical talent folk are not speaking up or clearly enough, which is to be expected.

Even if you ask then remind them of that, 10mins later you'll have to do it again, but try it.
If you have a good example of that you might try playing it to them, but nicely.

Doing aeroplane instructional vids, we get that, the occasional problem is, once the new talent get the idea, you're finished and onto working with new folk.
Our technique is to judge whether the talent is nervous about public speaking, some experts are, demonstrate the technique, then play them a *good* sample of their work, smiling the whole time.

Tech wise, reduce the low end at the recording stage and study where the best place to pin the lav is on their smocks.

Cheers.

Re: Basic suggestions for EQ on talking heads of various quality
 

Ditto the above, re: mixing on headphones and especially on 7506s.

As a starting point for EQ, I'd look at a spectral view of the air conditioning noise (on a raw track), to see where the problems are. Then roll off the bottom end to get rid of as much a/c as possible, but being sure that the rolloff starts no higher than 250 Hz. (This might make the voices slightly thin, but you're aiming for intelligibility.) Something not too sharp, around 12dB or 18dB per octave, would be a reasonable starting point... don't go sharper if you can avoid it.

After you've rolled off the LF a/c noise, you might want to use a little active NR to get rid of other background, but use it very carefully or you will make the voices sound robotic. The trick is to remove as much noise with fixed bandpass (or notch) filtering, before applying active NR.

No other generic suggestions... If you take the time to post a raw sample, someone might have more specific ideas.

Re: Basic suggestions for EQ on talking heads of various quality
 

Thanks guys. The quality is all over the place so posting a sample, I don't think will help too much. Also, because of the environment that I'm editing in - a cubical office - speaker monitors are not really an option.

All the advise you guys have given is great as I'm still working my way around the audio field. I guess also some suggestions for eq'ing the voice for clarity and intelligibility would be great. I cut off the low end a bit, drop the mid range as much as possible, and tweek the high end to taste. That's all I got.


-denez

Re: Basic suggestions for EQ on talking heads of various quality
 

One more thing. I can't for the life of me get a grip on the "compressor" filter (either single band or multi band). Instead I've just been using the "hard limiter" withe the ceiling set to -6db, and a healthy input boost.

I just discovered the "Dynamics Processing" and I'm really into it as I can selectively boost audio in different db ranges. The only problem is that I'm getting a sort of high-end "rustling", that sounds similar to the sound that a snare drum can make when is rustles in the background. De-esser helps a little. I don't think this sound is in the original recording either, but rather an artifact of all my tweaking.

Any suggestions?

-denez

Re: Basic suggestions for EQ on talking heads of various quality
 

Well, that will certainly reduce the intelligibility.

I respectfully still suggest you post one or more samples, so we can hear what you're hearing and what you're doing.

Re: Basic suggestions for EQ on talking heads of various quality
 

+1 Don't reduce the mid range, if anything, increase it around 2-5Kh .. about 2-3db. Search around the web for 'recording voices' or similar.

The 'rustling' sounds like you're not positioning your lav mic properly, it's picking up the sound of the speakers shirt etc.

Yes post some samples.

Cheers.

Re: Basic suggestions for EQ on talking heads of various quality
 

sheesh... always good to hear that your doing everything wrong....

Just kidding.

Anyways, here's an example of the "rustling," it's not the most extreme example, but it works. I also attached a screen shot of the "dynamic processing" that I used and that's all that's been done.

Let me know what you think and I'll defend my (good) intentions later.

-denez

Re: Basic suggestions for EQ on talking heads of various quality
 

for what it's worth, here's an example of the EQ'ing and other processing that I normally might do.

I did not do any noise reduction. Also, I'm not entirely happy with the results, but that is often the case. The voice sounds "fuller" to me, and dropping the mid range gets rid of either a "telephone-y" sound or a back of the throat "nnnngggg"-type sound.

I'm worried that I'm just butchering it and making bad audio worse...

Also, the hard limiter is giving an over all gain boost and clips at -6db. The de-esser is trying to get rid of some of that "rustle."


thanks again for all the help so far.
-denez

Re: Basic suggestions for EQ on talking heads of various quality
 

Thanks for posting these. The first 2 samples didn't sound too bad to me, I've heard a lot worse clothes rustle.

The sample above also doesn't sound too bad, don't forget viewers will adjust their playback volume to suit themselves.

This is not for transmission in a bunch of other over compressed soundtracks on TV, where yours might not sound loud enough in comparison. Is it?

Cheers.

Re: Basic suggestions for EQ on talking heads of various quality
 

Gotta be honest. It doesn't sound bad at all. The rustle that you describe kind of sounds less like clothing rustle than preamp or floor noise.
In any case it didn't sound bad on my computer but as Allan says, a lot will depend on the final tramsmission circuit.

Re: Basic suggestions for EQ on talking heads of various quality
 

I agree with Don, for the stated purpose I think the raw file sounds fine.

I hear absolutely no difference between sample #1 and sample #2, except that sample #2 is about 4dB louder than #1.

The room tone is not objectionable IMHO. It sounds as if it was recorded in a "real" room somewhere, and not on a sound stage. What do we see in the background of the shot? If we see a real working hospital room, then audio with a completely clean background would seem very artificial.

I don't hear an objectionable amount of air handling noise, which I expected based on your description. There is a bit of equipment noise, and you could lower that with some judicious notch filtering, followed by a small amount of active NR. You don't give us any significant amount of room tone, so it's hard to analyze to find the frequencies in question.

IMHO sample #3 has less intelligibility than the other two; that's because you lowered the upper midrange frequencies (around 5kHz) significantly. If you want to increase intelligibility, follow Allan Black's suggestions above.

I am not bothered by any clothing rustles.

That was not an objectionable quality in the track itself. That is the objectionable quality that is characteristic of the Sony phones. TERRIBLE phones for mixing, they give you a very inaccurate idea of what the track sounds like. I don't quite understand why you can't mix on speakers... somewhere, some time.

Indeed, a small amount of dynamic compression might help the track, it depends on playback circumstances. If it will be played on bad speakers in a noisy environment, use a little compression. If it will be played in a relatively quiet environment, on halfway decent speakers, the compression would be less necessary.

Re: Basic suggestions for EQ on talking heads of various quality
 

+11, Headphones are fine for acquisition...not for mixing/mastering decisions.
Doesn't sound too bad. I don't hear any really objectionable extraneous noise. If you listen extremely loud and/or with inappropriate gear, (common to those inexperienced in audio) you'll always hear something odd.

Re: Basic suggestions for EQ on talking heads of various quality
 

wow. Thanks so much for you guys working with me on this. It's looks like I'm being a bit anal, which I think is good quality for this kind of work.

The example I posted wasn't an example of one of the "bad" audio clips (believe me, I have some), but it just happened to be what I was working on at the time. The "rustling" I'm describing is not cloths rustle, but something that goes along with the vocals, maybe its preamp noise, it's in the high end, I know because I use the de-esser to get rid of most of it. I hear it in the "withfix.02" clip, so it's something the compression is emphasizing.

----------

Ok, lastly, any suggestions for good mixing headphone. The 7506's I'm happy with because they are industry stander (perhaps as field monitor, that's fine) and they only cost $100 or so, and I like Sony. Any good studio monitor phones in about the same price range?

Re: Basic suggestions for EQ on talking heads of various quality
 

There are NO good mixing headphones, period. Phones are fine for evaluation and editing but mixing should be done on properly calibrated monitor speakers. Headphones introduce their own coloration, particularly bass emphasis, as they interact with the wearer's head and mixes done using them don't often translate well to the normal loudspeaker environment in which the audience will usually be listening. For mixing you really need a decent pair of studio monitor speakers, calibrated for proper levels, and a work environment that has had some attention paid to acoustic conditioning. You don't have to go overboard but a workspace where some attention has been paid to the acoustics is essential.

Re: Basic suggestions for EQ on talking heads of various quality
 

I saw HPs in Mix (or other pro audio mag) that were supposed to 'accurately' simulate a speaker/room environment. My main recollection was they were extremely expensive.

Re: Basic suggestions for EQ on talking heads of various quality
 

When I'm recording music, I don't even like Sonys as field monitor phones. My favorite in the low-price/rugged/good-sounding range is the Sennheiser HD280 Pro. They don't have the screechy mids that the Sonys have... much smoother and gentler.

Re: Basic suggestions for EQ on talking heads of various quality
 

I also like the Senn HD280, I use them for music recording all the time, and sometimes for dialog. They are about the same price as the 7506. The Sonys still come out for dialog.

Do your edits on headphones, sure, there's a lot you can do in your cubicle with phones. Rick and Steve are 200% correct that headphones do not match most playback environments where your work will be heard.

Headphones mis-state EQ. They mis-state stereo image. The mis-state dynamics. This means that when you're mixing a couple simultaneous sources, you really have no guidance on relative loudness. As you start to work with compression, limiting, and dynamics, headphone monitoring is even worse!

We have this problem at my college - computer labs (not even cubicles!). When you have 24 students in a room, they've got to be on headphones. I can tell you from listening to hundreds of student projects that mixing on headphones is totally hit or miss. Simple tasks like VO over music or other background are totally inconsistent.

If you want to make your audio better, you have to hear what the audience is hearing. No ifs, ands, or buts. Do your final compression, eq & mixing after hours. Tell your boss you *must* book outside studio time for audio finishing. Send it to someone. Until your mixes are done on reference monitors you are working in the dark, your headphones are lying to you left and right!

This is basic to consistent quality in media production: You must see what your audience sees and hear what your audience hears.

Re: Basic suggestions for EQ on talking heads of various quality
 

You can "kind of" learn how to mix on headphones. One trick is to mix quiet things a bit too loud. If you try to be subtle, you will find that those sounds will disappear when played back on TV speakers.

I recommend listening to mixes on many systems (monitors, high-end stereo, TV, headphones, whatever) before settling on the final mix. Listen for problems like boomy or missing bass, an obnoxious instrument, or hard to understand dialog. Then come back to your system and make relative adjustments. Rather than have it sound great in your studio, make it sound acceptable everywhere. Most audience members don't have a clue about great audio - but they're audiophiles when it comes to hearing mistakes. :)