Needing some help on this project (Lecture speeches)

Hi, guys,
 
I have to clean 40 audio lectures (speech only) of about 2 hours - 170MB each, and they're in MP3 - 128kbit/s.  My questions are:
 
1- Is it better to convert the files to mono before cleaning?
2- I know that Sonar has a lot of resources to do the cleaning but I was wondering if Melodyne (stand alone) could do the task easier, even for big files (I didn't work with Melodyne yet and I purchased the editor edition), or you recommend another plugin inside Sonar? I really wanted something very straight forward for not loosing too much time in this project, the files are too big and there's some pops and cracks...
3- After the cleaning, would you recommend any effects or is it better to let them dry?
4- Since I'm gonna have to fit all the files in one 4.7GB DVD, what do you think is the minimum bitrate I can use to do the task without harming so much the audio?
 
Thanks a lot for any help!

You could probably get away with encoding at a lower bitrate than 128 for just raw speech.  We notice differences a lot more in musical applications than ordinary speech.  I would render test files out at 96 and as low as 64 and give them a listen.

@leebut
Sonar X3 does not have that built in, but  if you have X2, it came with R-mix which is capable of decent noise reduction. An FYI on using noise reduction plugins like the ones found in Audacity, Audition, and Sound Forge:
 
 1. Make sure you have enough of the actual noise you want to remove. Whenever I record sound on location, I record a good minute of the environment so I have a good variety of the type of noises that may be in a room.
 
2. Also, you will get less damaging results on applying noise reduction if you use modest settings.  Applying Auditions noise reduction twice at 50% will give much better results than once at 100% no matter how good the noise print is.
 
3. The more distinct the noise is, the better your results.  A beeping alarm clock is much easier to remove than background chatter.
 
4. If the noise is of the white noise variety that is hard to separate from what you are trying to record, skip the noise reductin plugin altogether.  You will just get a very tinny sounding voice.  Exceptions to that are things like forensic work where the goal is simply to extract something intelligible.
 
When I find myself in a situation where the white noise or chatter is too much to remove using noise reduction, I usually use a combo of filters and gates.  Be careful with noise gate beccause it can sound odd if you have background noise during speech and dead silence otherwise.  A way to fill in the silence but at a level that is controllable in ratio to your main speaker is to make a loop of the ambient recording.  on another track in your project.  It will make the sound between the gated parts sound more natural
 
 
 
 

Do you have melodyne editor?  The version that comes with Sonar will not process polyphonic material.  I doubt even Editor can do the task you speak of. but let us know if that works. Converting to mono will reduce the file size.  I doubt you will run into phase issues.

dubdisciple,
 
Yes, I have the editor, I benefitted from the promotion.
One of my concerns in converting the files to mono was related to phase issues but you kind of allay me.
 
Thanks again!

LunaTech,
 
certainly, RX3 is "The" software to do this work but, after Sonar X3 and Melodyne Editor, my budget are low to spend more US$349, for the basic version; the advanced is US$1,199!
 
Thanks for the reply!

I was using the spectral tools in audition for this, but I find myself using R-mix more.  Audition gives me slightly better results, but R-mix is a quicker workflow.  I don't have editor, but would be curious about the results.

John,
 
the goal is to put all the 40 files (more than 60 hours of lectures) in a 4.7GB DVD.  This project will be a part of an audio book and the autors need to save money, so that, double layer DVDs aren't an option, they're still expensive here...  I've already received the files in MP3 format and it's going to be the final format, otherwise, they woudn't fit on the DVD.
 
Thanks!

For an overview of basic R-Mix operation check out the video here

Hi, Scook, thanks for the input.

Atsuko
Sanderxpander, maybe Melodyne can do the job but, as I stated above, for long files it gets too much of the machine making it almost impossible to work.  Could you give me a more detailed explanation how to do the phase thing? Thanks!

As I said, Melodyne is entirely the wrong tool for the job. It does pitch and timing corrections mainly, it wasn't made for speech at all unless you were planning on slowing down or speeding up certain words, or raising or lowering the pitch of the voice.

The phase thing really depends on the kind of noise, basically you try to find a long bit of static noise (like, no pops or crackles, but hum or white noise). Copy it to a second track and flip the phase using the phase button on the console strip. Now listen and see if it cancels out the noise. If it does, try it under different parts of the thing. If it's static enough, you may get a significant improvement. If you do you can try copying it under different parts, but you'll have to pay attention that the noise/hum part of the waveform lines up. I've gotten rid of some hum using this in the past, but honestly, if you have no idea what I'm talking about or don't know how to go about it, it's not really worth it because it's a pretty finicky process and the results may be disappointing.

We are all here to help each other.  I have received plenty of help on here.  It may help if you post a tiny segment of exactly what you are dealing with. I understand privacy and copyright issues, but perhaps enough to not violate either.  Just a sentence or so.  If that is impractical, I understand.  The phase trick described by another poster is often used by remixers to obtain acapellas from songs.  Worth a try and you know right away if it worked.

dubdisciple, 
 
thank you very much!!  I'm at work now, I'll do it later when I get home.
 
Thanks again, this forum is awesome!

Sanderxpander
Atsuko, yeah you got the right idea with the phase thing. It's worth a try at least. You'll know if it does something within a few minutes.

There are also a few free VSTs though I have no idea how good they are, such as this one.
http://www.synthtopia.com...-software-for-windows/

And Audacity has noise removal of course.
Alternatively, there are noise reduction plugs for any budget really. Voxengo makes one for just 100 bucks. And plugindiscounts.com offers RX3 for 230 only. And a Waves plugin specifically for dialog tracks for 170.

Depending on how recurring this kind of job is and what (if anything) you can bill the client it might be worth getting a specialized tool.

Now that is a thought. You may be able to for a fee subscribe to adobe for a month (of for however long it's needed), use the software (Audacity) to do the work and be done at a minimum expense.  I also completely agree that if this a recurring job or even a new skill set that you can now offer as a service.... A dedicated tool will pay for itself in a short time. Good Luck..

Sanderxpander
Why would you subscribe to Adobe exactly?

To use audacity from their cloud service as opposed to paying full price for the program.  I believe that you can subscribe to the cloud for full access to Adobe's programs for a given period of time and then you are done.