Warbird wrote:
Hello,
I'm very new to this forum and have looked around. Looks like some pretty smart people so I figured I'd sign up, watch was going on and tyr to learn.
I'm very new to this whole audio editing thing and know very little about it.
------------------------------------------------------
You are hardly alone!! Welcome to The Club.
------------------------------------------------------
Here is my question:
I have a wav file, 8000 bit mono, that I need to cleanup. I am looking to try to bring the voices from the background forward to make them understandable and remove or reduce the front noise. Most of the noise that drowns out the voice is loud music, TV, car and a subway train etc. I have downloaded Wavepad to play around with but I quickly figured out that I really don't know what I'm doing.
Can anyone help me to get this recording cleaned up?
I hope I can do this, the recording was made over two days.
Thank you in advance
------------------------------------------------------------------------------------------------------------------------------------------
Assuming that the background voices and the noise that is drowning out these voices _occur simultaneously_, the principle that is involved is to attempt to differentiate between the noise and the voices by means of frequency discrimination. That is, if the voice and the noise occupy primarily different portions of the audio frequency spectrum, then it will be possible to achieve some significant amount of noise rejection by resorting to a frequency filter which is "tuned", so to speak, to the voices that you want to hear, while tuning out the noise.
To this end, WavePad has an adjustable noise filtering mechanism, by which you can play around with the setting of the frequency band which you wish to keep (or reject). First, you might try a simple low-pass filter, which passes through (without significant attenuation) all frequencies below a certain "cutoff" frequency that you can adjust. WavePad's noise-rejection filtering system permits you to adjust the frequency response curve (a plot of frequency versus amplitude). I would try a cutoff frequency at about 3000 hz, apply this filtering to your selected portion of the audio file (.wav) and then listen to the result to see if the voices that you want to preserve are sufficiently intelligible (that is, have enough high-frequency content). Then see if the offending noise has been reduced to an acceptable amount.
It may not, of course, since the frequency content of the noise may intrude into the portion of the frequency spectrum that you have kept in order to make the voices intelligible. If the voices are intelligible but the noise is still unacceptably high, then try reducing the cutoff point of the low-pass filter, perhaps from 3000 hz down to 2000 hz, and listen again. But before you make this change, be sure to go to the Edit menu and undo the filtering that was not "successful". Then, after making the change in the filter, listen again and see if the voices are still acceptably intelligible, etc.
As long as the voices are acceptably intelligible, keep reducing the cutoff frequency of the low-pass filter until you reach a point where the voice quality is degraded to an unacceptable point; i.e., with too much of the high-frequency content removed. You will have to make this judgment yourself--no one else can make it for you. In other words, the point where you stop (i.e., do not feel that you choose a yet lower cutoff frequency) is a compromise (sigh!) between the quality of the voices and the reduction of the background noise.
Got the picture?
Play around with it and see if you can do very much with it. As you can guess from what I have discussed above, there are no guarantees :-( Every case is different.
Musikone