 |
VO-BB - 20 YEARS OLD!
Established November 10, 2004
|
| View previous topic :: View next topic |
| Author |
Message |
D Voice
Been Here Awhile
Joined: 26 Jun 2010
Posts: 232
|
 Posted: Thu Mar 08, 2012 8:37 pm Post subject: Noise Reduction sampling |
 |
|
On thing that has always bugged me is getting audio software noise reduction right: aggressive enough to remove unavoidable background and self-noise, yet without noticeably dampening the sound, or getting those horrible metallic or whirring sounds (Adobe Audition, I am looking at you!). Audio software and plug-in creators all state the obvious- about choosing a "quiet" passage to sample that is "long enough", while being disappointingly vague regarding how long a sample is too short or too long.
It seems I asked this somewhere before on a forum and got the counter-intuitive answer not to take too long a sample- about 1 second or so was sufficient.
[counterintuitive because it would seem to me that the noise reduction algorithm would average out the various frequencies over the sample period, and a longer sample would be able to more properly dismiss any rogue transient sounds that might have slipped through]. Does anyone know for sure?
Or at least for the purpose of friendly discussion, among the VO artists and engineers here, how about Noise Reduction software: what is your experience with sampling lengths? Then more broadly, perhaps people might share regarding settings for the Noise Reduction on their various software (realizing of course, that every noise floor situation will be somewhat different).
Last edited by D Voice on Thu Mar 08, 2012 10:51 pm; edited 2 times in total |
|
| Back to top |
|
 |
D Voice
Been Here Awhile
Joined: 26 Jun 2010
Posts: 232
|
| Posted: Thu Mar 08, 2012 8:38 pm Post subject: i was reminded of this issue because... |
|
|
I came across this the other day:
| Quote: | Pro Techniques from VH-1's Behind The Music
Washington, D.C., mixer/editor Richard Gray suggests finding the dialog-free room tones...and using X-Noise to sample and "learn" those ambient sections of noise. The larger the sampled area, the more effective the plug-in can be in processing the audio clip and removing unwanted noise from one's critical dialog tracks.
"Since I usually have at least five-second handles on either side of the clips from the OMF of the session, I have more opportunity to find a larger section from which to sample some dialog-free room tone before the interviewee starts speaking," Gray continues. "Once I've sampled that background noise I can use the same setting on other clips from the same interview throughout the show."
For a good starting point in most dialog situations, he suggests setting X-Noise's Threshold parameter between 12 and 17, adjusting its Amount of Noise Reduction control no higher than 70, and setting X-Noise's Attack setting in the mid-20s and its Release parameter in the mid-200 range. |
|
|
| Back to top |
|
|
Benjamin Stovall
Been Here Awhile
Joined: 13 Dec 2011
Posts: 250
Location: Los Angeles, CA
|
| Posted: Thu Mar 08, 2012 10:22 pm Post subject: |
|
|
I'm pretty finicky when it comes to noise reduction. I really despise that over-processed sound (or "warble" as I refer to it) that you can get if you're not judicious with it.
In my experience it has worked best when there wasn't a significant amount of noise to remove in the first place. If I'm just cleaning up a slight offense, I have gotten some very good results--but some software works better than others.
I have a hard time getting good results with Audition's noise reduction, personally. More often than not when I treat audio in Audition that has more than a little noise to reduce I get the "warble."
I soon realized that to reduce the noise sufficiently I had to get fairly aggressive with the reduction percentage which exacerbated the problem. Oddly enough, in a pinch I used Camtasia (it's a program aimed at screen capture) which has nearly zero options for audio cleanup other than creating a basic noise print and got MUCH better results with only a one second or so selection for the noise print. Go figure. I think it may all be in how good the algorithm is.
At this point I've tried quite a few (including X-noise) but, my best experiences have come from Izotope's offerings, with a small or large selection for "learning" yielding about the same results to my ears.
_________________
Ben Stovall Voiceover
http://www.benstovall.com
"When you're nearing the end of your rope, tie a knot. Keep on hanging. Keep on remembering, that there ain't nobody bad like you." -- The Electrifying Mojo |
|
| Back to top |
|
|
Frank F
Fat, Old, and Sassy
Joined: 10 Nov 2004
Posts: 4421
Location: Park City, Utah
|
| Posted: Thu Mar 08, 2012 11:33 pm Post subject: |
|
|
| Quote: | | Benjamin Stovall writes: :...it (noise reduction - processing) has worked best when there wasn't a significant amount of noise to remove in the first place. |
This comment is almost on track. The first step in noise reduction is the room, then the mic, the voice, the pre-amp, the sound card, and then the editing software are the culprits.
I have made a specific comment often regarding processing and I will again state it here: ... a file should leave your studio with as little processing as possible and your recording chain should ALWAYS be as clean as possible - period.
Please remember: the client cannot un-process your audio. In some cases - very rare cases such as you are sending the file to radio or to IVR a (very, very, very) little processing may be required; but in every other case sending a clean, un-processed, flat file is the best representation of your proficiency in recording. IF ANY PROCESSING NEEDS BE DONE, LET THE CLIENT DO IT UNLESS OTHERWISE REQUESTED.
I hope that makes things perfectly muddy.
Frank F
_________________
Be thankful for the bad things in life. They opened your eyes to the good things you weren't paying attention to before. email: thevoice@usa.com |
|
| Back to top |
|
|
Marik
Joined: 31 Jan 2012
Posts: 9
Location: SLC
|
| Posted: Fri Mar 09, 2012 12:09 am Post subject: |
|
|
It is not easy to find a good NR with minimum artifacts. Even harder thing is to learn how to use it correctly and find good settings for each individual case, which takes lots of trial and error. So far for me Izotope worked the best. Not exactly cheap, but with educational discount can be had for half price. I used it even on commercial CDs with good results.
Best, M
_________________
Mark Fouxman
Samar Audio & Microphone Design
www.samaraudiodesign.com
The Art of Ribbon Microphones--design, repairs, re-ribboning, modifications, motor machining and fabrication, transformers, and more... |
|
| Back to top |
|
|
georgethetech
The Gates of Troy
Joined: 18 Mar 2007
Posts: 1878
Location: Topanga, CA
|
| Posted: Fri Mar 09, 2012 12:13 am Post subject: |
|
|
You just gotta use your ears when setting these things.
Most algorithms are pretty mediocre, until you get into stuff like "CEDAR".
I find a well tuned noise-gate far better in almost all cases.
_________________
If it sounds good, it is good.
George Whittam
GeorgeThe.Tech
424-226-8528
VOBS.TV Co-host
TheProAudioSuite.com Co-host
TriBooth.com Co-founder |
|
| Back to top |
|
|
Benjamin Stovall
Been Here Awhile
Joined: 13 Dec 2011
Posts: 250
Location: Los Angeles, CA
|
| Posted: Fri Mar 09, 2012 12:17 am Post subject: |
|
|
| Quote: | | IF ANY PROCESSING NEEDS BE DONE, LET THE CLIENT DO IT UNLESS OTHERWISE REQUESTED. |
I agree with that sentiment wholeheartedly, though in some folks situations it's not always viable. If you find yourself in the spot where your audio isn't clean enough to send to a client, but it HAS to get sent, you may need to use some noise reduction. If you send out a noisy file to a client it may be your last with that client. If they need to compress it to raise the level to cut through the mix for example, the noise will get even more pronounced. And more importantly, that's not the kind of processing you want your clients to have to do. Most clients expect (and deserve) audio from you that is at usable noise ratios.
If your chain/room are producing too much noise, NR could be a short term fix out of sheer necessity but it should NOT be noticeable or its moot. You need to get things sorted ASAP if this is your situation.
Also, there are a few successful folks I know who use it quite often to remove SLIGHT noise before sending to clients, and the reduction is just as slight so you would never know it was done. To me, that's the rule: If you can hear the processing it's too much, but processing in and of itself is not anathema.
EDIT: I wanted to comment on this: | Quote: | | I find a well tuned noise-gate far better in almost all cases. |
What I sometimes run into with noise gates is if the noise is still layered under the speaking parts I can hear the difference from the silences no matter how much I play with the attack and release times. Perhaps the kind of tuning you are referring to can even deal with that. If so, it's something I'd like to learn how to do.
_________________
Ben Stovall Voiceover
http://www.benstovall.com
"When you're nearing the end of your rope, tie a knot. Keep on hanging. Keep on remembering, that there ain't nobody bad like you." -- The Electrifying Mojo |
|
| Back to top |
|
|
D Voice
Been Here Awhile
Joined: 26 Jun 2010
Posts: 232
|
| Posted: Fri Mar 09, 2012 12:40 am Post subject: |
|
|
OK, maybe I didn't make it clear enough. OBVIOUSLY, the best way to reduce noise is to eliminate as much as possible at the source- gear, room, isolation, acoustics, etc. And just as obviously, if the client is a pro engineer and wants something totally dry and is better qualified and equipped to handle all the processing, etc. by all means, it is better to let them do it.
But in spite of that, in my case (and perhaps in the case of others as well) there are various and numerous situations (including auditions, small-scale video producers, events, posting stuff, etc.) where something does need to be done in the way of noise reduction, and we/I, as the VO person generating the audio file, is the person who needs handle it.
SO...in such cases, I would like shared advice, comments, research, etc, about doing a better job with NR, without dumbing it down too much. |
|
| Back to top |
|
|
sounddguy
Contributor IV
Joined: 22 Jan 2009
Posts: 100
Location: Atlanta, GA USA
|
| Posted: Sun Mar 11, 2012 6:00 pm Post subject: Re: Noise Reduction sampling |
|
|
| D Voice wrote: | | ... aggressive enough to remove unavoidable background and self-noise, yet without noticeably dampening the sound... |
All processing is like adding spice when cooking, a little bit goes
a long way. You may have better luck by making several less aggressive passes, re-sampling the noise each time. Use your ears between
each and go back to the previous pass when you start to hear artifacts. |
|
| Back to top |
|
|
|
|
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum
|
Powered by phpBB © 2001, 2005 phpBB Group
|
FAQ
Search
Memberlist
Usergroups
Register
Profile
Log in to check your private messages
Log in
