I’ve read about problems with Noise Removal, such as artifacts showing up. That is not the problem I am encountering. On seeming random occasions it simply does not work, even when using it in the same file. I’ll capture a noise sample and then apply it to a copy of a raw track and get results and then I’ll apply the same removal to another copy of the same original track with only a slight change, such as an extra 2 dB, and it does absolutely nothing. Has anyone encountered this. AFter that even outrageous changes to the setting, like removing 25 dB, does nothing. I’ve tried rebooting but sometimes it works and sometimes it doesn’t. Quite stymied here.
Since both noise reduction and file management are wildly and broadly misunderstood, we need to get down much finer than “my file sometimes doesn’t work.” What precisely is the work; what are you removing and from what?
We warn people not to do production in MP3 or any other compressed delivery or internet formats because the music after is always different from the music before. Almost all compressed formats cause damage and although they may sound the same, if you get the profile from one and try to reduce another, it may not work. I believe the instructions are clear the profile needs to come from the exact same show as the reduction. We even get nervous when people claim they’re taking the profile from the beginning hour of the show and applying it to the second hour of the same show – all else being equal.
Koz
Koz,
The files I am dealing with are from live recordings of our concerts.
The files are recorded in wav format 44.1 kHz - 16 bit
We are not using sophisticated recording equipment - a Zoom Handy Recorder H4n
We determined, by recording silence at a recording level similar to what we set at the concerts (60 to 65) that there is some inherent white noise being captured in the recording from the recording unit itself.
Surmising that this noise is going to be in ALL of our recordings we have been trying to remove that specific noise. Generally it’s only a notable problem at the beginnings and endings of recordings as we play a lot of music which begins and ends very quietly.
We have also tried capturing the “noise” in the hall at actual concert, 2 to 3 seconds of the noise just prior to the beginning of the music we are applying the noise removal to.
The frustration is that “sometimes” it has worked and “sometimes” it has not with both “captured noise” samples noted above.
I began getting frustrated when I had applied the noise removal to a specific piece, liked the result, and then decided I might be able to remove even more noise. So I undid the noise removal, went back into the Noise Removal settings and made minor tweaks to the settings. An example was to change the dB from 8 to 10 - not a gargantuan change.
Then when I applied the noise removal to the very same track in its original unedited form with the very same captured noise with that one small change “nothing” happened to the music. No noise was removed whatsoever - at least that we could hear.
This has happened to us several ties now with several tracks.
On other occasions it has worked fine. I can’t see a pattern here yet.
For best results you need to grab a separate noise profile for each Audacity track you want to remove noise from. The “white noise” generated by the hardware may be subtly different from recording to recording and also has different ambient noise mixed in for each case.
To see what frequencies are actually in the noise, select a region that has noise only and look at it in Analyze > Plot Spectrum.
I don’t see on the face of it why removal at one dB setting would work and a slightly higher setting wouldn’t, but it’s impossible to comment without the actual audio you are using. Please export a WAV that includes some noise on its own and also what you recorded that has the same noise in the background, zip it then upload the zip to a free file transfer service such as:
http://www.yousendit.com
http://www.sendspace.com
http://minus.com/
Then give us the web address where we can download it.
Gale
These are the fun ones. You’re doing everything perfectly and exactly correctly in every way…and it doesn’t work. That just means we’re all looking in the wrong place.
I gotta leave for a couple of hours. I’ll read over this again when I get back.
Koz
apply it to a copy of a raw track
This is where we start getting very particular about terminology. You are not using “raw” format sound.
The files are recorded in wav format 44.1 kHz - 16 bit
You’re perfectly correct that the background noise of the Zoom, the white, rain in the trees, hiss, is going to be common to all the recordings, or very close to it. You’re not correct that you can get rid of it. Noise removal works by internally generating a suite of filters and notches appropriate to the profile. If you apply Noise Removal at, say -18 to the exact place you took the profile, that portion of the show should vanish. Repeatedly. All the other NR settings are zero.
You might try that.
The problem with white noise is it contains all frequencies. That means with no other adjustments, settings or modifications, Noise Reduction will try to suck the whole show out of the show. This isn’t desirable, so the Smoothing adjustment is provided. What that does is recognize valuable portions of the show and stops noise reduction during those portions. The is the Dolby Philosophy that you can’t hear noise during actual musical notes and voices.
What happens when you try that above test?
Koz