Using MTU DNoise V3.0

[W. Wheatley]

OK, so I\'ve taken a sample of noise- a nasty \'hum\' caused by bad mic. hookup- and I make a template, which I want to apply to an entire file. I\'ve gotten that far. The file recording is speech only- a lecture- the \'hum\' is very loud- probably 1/2 as loud as the speakers voice.
Please give me some hints as to hoe I could set this up, with specific settings for all parameters. I\'m a bit of a novice at this, and even though I\'ve read thru the 15 page Introductory Guide, I\'m getting horrible results, which I know are only due to the fact that I don\'t know what I\'m doing yet.

[admin]

Noise removal is an \"art\". It takes trial an error to tweak the parameters for the best processing. Because DNoise V3 allows live adjustment of the noise parameters, you can quickly start to see the interactions of each control on the particular audio you are processing. Each audio file is unique! You can find some similarities - for example removing cassette tape hiss from a sermon from tapes made at the same church every Sunday. You will also be removing some ambience noise of that church, so the same template will not work as well on tapes from another church, with different room ambience.

I strongly suggest you read the manual before starting as it has many tips that will help you understand.

Make sure your noise template is in the range of 1/4 to 1 second long. This can affect your results.

Don\'t start with the WORST CASE you have ever run across! Pick a really simple noise case (like flourescent hum) not removing air conditioner noise. If you try the worst to start with you will become discouraged. Experts can struggle with removing some noises. You need to try to understand the basics first, then move on to advanced noise removal.

[geezer]

---Although I have yet to use the newest version of Dnoise, I had great success with the earlier versions, and some techniques seem to apply to most de-noising software:

---You will often have great success with multiple passes- i.e., don't try to get all of the noise out on one pass. You can adjust the settings and get new templates on each successive pass, and you will tend to have fewer nasty, side-effect artifacts in the final product.

---Sometimes a shorter template is better than a longer one. 1/4 second seems to be the minimum length, but longer is not always better. The "character" of the template matters, as well. Try several. You have the luxury of real-time processing now. In "the old days", we would have to do considerable thumb-twiddling while we waited for our test paramaters to process (originally 27 times real time!)

Good Luck.

[arminroesch]

Originally posted by W. Wheatley
OK, so I\'ve taken a sample of noise- a nasty \'hum\' caused by bad mic. hookup- and I make a template, which I want to apply to an entire file. I\'ve gotten that far. The file recording is speech only- a lecture- the \'hum\' is very loud- probably 1/2 as loud as the speakers voice.

In that case I suggest to use the EFX Filter in MicroEditor to filter the entire file with a 300Hz Highpass function. That will give you a much better S/N by a easy way.
Regards

[Hessel]

My stereo files sounds narrow after a ligth processing with
DNoise 3.21. I use only a pair of precision mics and pre to record acoustic music. The sound is good but the image is very poor after the processing.


Hessel