A friend has asked me try and "fix" some audio tracks for a talk show she is producing for internet distribution.  I am a musician and composer who knows a little about mixing music, less about post-production sound, and virtually nothing about mastering.  I'm hoping some of you experts out there can give me some advice.
 
I have three DAWs I can choose from -  Sonar, Logic and Pro Tools.  My mixing tools/plug-ins/effects include the standard stuff included with these DAWs:  EQ, compression, multi-band compression, limiting, etc.
 
I've warned the producer that this is a "garbage-in/garbage-out" situation, the original audio was poorly recorded and has a lot of problems.
 
First, there are several places where the close mics cut out, leaving only the camera mics.  The audio from camera mics is clearly more muffled and has a lot more room reverb.  I've been asked to "fix this and make it all sound homogeneous" (!)  This seems to me like a mastering issue, and not something that can be fixed with my basic mixing tools above -- if it can be fixed at all.  Is there anything I can try using those basic tools?  Are there any mastering tools out there that a noob might be able to use to address this problem? 
 
Another issue is the hiss on most of the tracks.  I suspect this comes from a noisy mic preamp turned up too high.  The hiss is too broad-spectrum to treat with a notch filter.  I tried using a lowpass filter and I found that a cutoff point around 3.5k eliminates a good bit of the hiss without degrading the quality of the dialog too much.  Is there another approach that might be better?
 
Finally, on some episodes there is a noticeable difference in level between the host's voice and the guest's voice.  Both voices were recorded to the same track, so I can't just tweak the track levels.  I've used some basic compression to smooth things out somewhat, although you can still hear some difference between the two people talking.  Is there another approach to the problem?  Can someone give me some tips on the best way to use compression to deal with this specific issue?
 
Any advice you can give will be truly appreciated.  

Do you mean to say that there is more than one episode with bad sound?
 
I suspect you may do an enormous amount of work to get results that the "producer" dismisses as disappointing.
 
There was a recent thread in the software section about tools that de-reverberate signals. That type of tool might help with matching the camera mics with the missing lavaliere audio.
 
There are a number of tools for filtering out noise. The cheap noise filters require tedious craftsmanship and a lot of attention to detail, but they work OK.
 
The expensive noise filters work pretty good but cost $10k to $15k. The makers can get that kind of money because sometimes a very expensive film production can't enjoy the luxury of a quiet setting so the seemingly high cost is compared to all the other costs and the impossibility of capturing sound some other way.
 
You should probably decide if you really want to start this project.
 
My day job is audio for field production video... all of the problems you describe should have been avoided or managed and any producer with more than one episode under their belt should have facilitated solutions either with proper staffing or appropriate technology to make sure that the corrections were made before moving on.
 
I'm sure you can do some good work but it's hard to think you will find it enjoyable.
 
Good luck.
 
wishing you the best,
mike
 
 
 
 

It would have to be a very good friend before I'd even consider going near that project. You're looking at a lot of detail work that even an experienced post guy might have trouble bringing up to broadcast standards.
 
First thing I'd do is separate the two parts - with and without the main mic - into their own tracks. You'll have to EQ them separately in order to get them sounding similar. There are spectrum-matching plugins for just for that purpose, but they cost money. If there's no budget, you can still get there with the help of a good spectrum analyzer.
 
Next I'd apply volume automation to get the two voices to approximately the same level, and then compress the daylights out of everything.
 
There are tools that do a great job of hiss removal. You just have to give them a sample of the hiss in isolation for it to create a sonic signature from, and then let it go. No such tools come standard with any DAW that I know of, but the free Audacity editor has one. I haven't used that one myself, but I use Adobe Audition for such things and it does a remarkably good job of cleaning up such noise.
 
For de-reverbing you can try a gate and see how that goes. Another option is a transient shaper. SONAR's transient designer isn't very good but maybe Logic has a better one. Or if the producer is willing to foot the bill for a cheap plugin, grab either Meldaproduction's MTransient or eaReckon's TransReckon plugin. Both are currently on sale and both may be automated so that you're only suppressing ambiance during the parts where the main microphone has cut out.
 
No matter what, you're looking at a lot of tedious hand automation. Make sure the producer knows how much time you've put into it, because it's going to be substantial.

If it was recorded poorly, that will be the sound no matter what.  Enhancement for speech is a time consuming and complicated process requiring special tools.   If you want to invest in some you can charge your friend $5000 for the service which is not out of the question for forensic restoration.  It will still sound bad after, but at least it would be intelligible.  
 
You were right to tell him "garbage in/garbage out."  Polishing turds is a waste of effort.  Maybe next time he will spend the necessary resources to get better audio.  

I can't add to what has already been said. It's a hard job to fix what was bad to start with.
 
This should be a lesson learned by your friend. Get it right the first time and you don't have these sort of problems.
 
I'd consider the first attempt a loss and simply move forward avoiding those mistakes again.
 
Help them attempt a fix if they need the first episodes for "posterity" and let it go at that.  It won't be anywhere close to broadcast quality but it is what it is.

mike_mccue
Do you mean to say that there is more than one episode with bad sound?
 
I suspect you may do an enormous amount of work to get results that the "producer" dismisses as disappointing.
 
There was a recent thread in the software section about tools that de-reverberate signals. That type of tool might help with matching the camera mics with the missing lavaliere audio.
 
There are a number of tools for filtering out noise. The cheap noise filters require tedious craftsmanship and a lot of attention to detail, but they work OK.
 
The expensive noise filters work pretty good but cost $10k to $15k. The makers can get that kind of money because sometimes a very expensive film production can't enjoy the luxury of a quiet setting so the seemingly high cost is compared to all the other costs and the impossibility of capturing sound some other way.
 
You should probably decide if you really want to start this project.
 
My day job is audio for field production video... all of the problems you describe should have been avoided or managed and any producer with more than one episode under their belt should have facilitated solutions either with proper staffing or appropriate technology to make sure that the corrections were made before moving on.
 
I'm sure you can do some good work but it's hard to think you will find it enjoyable.
 
Good luck.
 
wishing you the best,
mike
 
 
 
 

Mike,
What is your opinion of Izotope's RX3 plug-in (base and advanced)?  From their website it sounds like it might be useful in fixing some of these issues (may still require a lot of work).
 
I am weighing purchasing this or Adobe Audition to handle post production audio editing task and wonder which one will be more effective.
 
Sorry to sort of hijack the thread, but the question is related and maybe the answer can help.

What is your opinion of Izotope's RX3 plug-in (base and advanced)?  From their website it sounds like it might be useful in fixing some of these issues (may still require a lot of work).
 
I am weighing purchasing this or Adobe Audition to handle post production audio editing task and wonder which one will be more effective.

 
I use both Audition and RX3 advanced.  The tools in Audition are not nearly as advanced as RX, the latter is pretty much as good as it gets in the restoration world.  I only use Audition noise reduction tools for quick background noise elimination for radio broadcast.  
 
At any rate, like I said.  It is best for removing troublesome noise from already great sounding audio.  It won't make bad sound sound better.

Hi Dave,
 I wish I had an answer for you. I have often wondered how good it is.
 
 I use Adobe Audition's basic offline tools on the occasions I do this work at my place. I also have access to CEDAR hardware up in town for the rare occasion I need it.
 
 I think the math in Audition and RX3 is very similar but that RX3 has the real time and other time saving options.
 
 I specialize in field work so most of my stuff is sent to post production studios in NYC, Bristol Conn, and L.A.  I don't know what tools they use at those facilites. Having said that we surely try to minimize the need for such tools and so we don't hear much about it unless it comes up in CYA discussion with the producer just before we roll camera in a adversarial situation where the sound will definitely need some cleaning up.
 
 When I occasionally do post for small video projects at my place I wait until near the end to do noise removal and so the relatively long time t takes to do it with Audition is focused on the shortest amount of material. That work flow works ok for me.
 
 If I needed to do this sort of work more often I'd definitely check out RX3 and try to demo it intently.
 
 I'm sorry I don't have a real answer for you.
 
 

"RX3 advanced... ...is pretty much as good as it gets in the restoration world."
 
I don't want to argue with Daniel but I think there is probably a very good reason why CEDAR hasn't had a need to discount their stuff so it competes on price with RX3. That stuff is black magic.
 
:-)
 

Mike, thanks for the insight above.
 
I currently do the field recording for a small video production company.  Right now, our video editor does the noise reduction and leveling using Audition on his video editing rig using Adobe.  I'd like to take some of that burden off him (and leverage my audio expertise) which is why I was inquiring about which tools to use.
 
Comparing Audition's subscription cost to RX3, I think RX3 is a much better deal over time, so I think I will be checking out the demo on the next project we do.