HELP removing background noise!
I'm a student, I'm working on a video project. The video was shot in a large room with the aircon on, hence the microphone picked up a lot of background noise. I'm new to audio editing and I've tried adobe audition to remove the noise. I've had a bit of success but the edited audio has a echo, the volume is reduced and there is a strong emphasis on the letter "S".
I am posting a small sample of the audio, I was wondering if you any of you would be able to edit it for me so that I know what settings/filters/processes I can use to achieve the best sound.
Sample Audio: http://rapidshare.com/files/69541725/UL_Consent.wav.html
I would be very grateful for your help.
You'll probably get a few people willing to help with
this - I'll look at it over the weekend, too.
The best noise reduction is amplitude dropouts, most especially with non repeating noise. Why is simple enough. if you repeatedly drop your amplitude using i/o signal reduction, you will eventually get your noise to near zero or zero signal. the downside? as you boost back to your nominal level you will undoubtedly add repeating signal noise. this is removable. just record dead air through the same pass through for 12 seconds, save the file, and use that wav to create a noise reduction pattern. runthe noise reduction on the file you created from your pass through and it will clean up nicely. this is an old radar signal technique adapted to sound. in radar, theres a law af averages that is an adjustment to the whole thing, as certain exactness isnt required. sound is different. if you record the same amount of dead air as your audio then phase ityou can effectively wipe it out, but you might drop some sound values. the moise reduction works better, as it can calculate those differences using an accurate balance.
Also, In light of recent problems found with audition, consider the following:
Make a Batch file in windows that will
1. take a dropped file as input(your session file will do)
2. get the parent folder (the whole structure is important)--this is the folder where your session is, along with the folder containing your audio files that you record.
3. periodically non-destructively (as in not deleting files that are no longer in the first folder) Mirror the contents of the parent folder to another location anywhere on any drive (every 2-5 minutes works) and make sure you set this folder, or you can simply use the parent folder of the original file and add a Backup folder inside,
4. then just continually make backups using a forever loop (Condition is always true like While 1==1). You can close it when you close audition by simply closing the Command prompt window. You won't lose everything even if you crash, after all, audition cannot delete what it knows nothing about.
On mac open automator:
start by making an application--call it BackupStart. create 2 Path variables: SessionFile, ParentFolder. These will be what you use to grab your files. Create more Path variables with names similar to BackupFolder#, where # is the number of which backup folder. You can now Set Variable Value (in actions library) for the path Session file as your first action (this will catch the path of the file you drop on the app). Next, run a shell script in Bash (an action in utilities or system), and set STDIN to Arguments. Clear everything in the shell, and type: dirname "$1" Exactly as shown (don't replace dirname, it's a command that grabs the name of the parent folder of the file you just dropped in. Add another Set Variable action for ParentFolder. Now use an ASK for finder items action, and look in it's options to "Ignore this items input" and check the box. Add a Set Variable for BackupFolder1. Repeat for each BackupFolder variable you have (ask for folder, set variable). Now add a GET VARIABLE action, go to options and select "Ignore input" again, make sure you are GETting the ParentFolder variable. Now another GET Variable for Backupfolder1, but DONT ignore input (you want the two to pass into one another and continue on). Repeat this last op for each BackupFolder variable, leaving the ignore input unchecked to group them all together. Now add a RUN WORKFLOW action and turn off "Wait for workflow to finish". Save this file, leave it open, and go to file Duplicate. Rename the duplicate BackupLoop1. File >Convert this doc to a Workflow. Delete the variable SessionFile from this document and all but the very last of the actions(Run Workflow). The other variables are still necessary. Everything we add should be above the RUN WorkFlow action. Use a Get Variable on your ParentFolder, and as before, do not check ignore input. You need this to run straight through from the first document. Add the GET Variable for your backups. Add a Shell Script in Bash, with STDIN set to Arguments. Clear the script box, and type:
rsync -vau "$1/" "$2/" (enter)
rsync -vau "$2/" "$3/" (enter)
The first line copies your parent folder's contents, the second copies the first backup to the second. You can continue this until you have handled every backup in the script. Apply a PAUSE action for a few seconds. Now add a Loop Action, and set it to run 50 or more times (applies a wait time until finish) and set it to use the same input. This will continually backup all your data as you record, and when you hit stop, you should get a copy of your audio almost immediately after, done by your system, and making the RAW file data into a finished file set. Now add the GET VARIABLE set again for all your variables, ignoring the input of the first one, but keeping it for the others. Point them into the last action of RUN WorkFlow. Again, Duplicate the document, call it BackupLoop2. Change the Run Workflow in this file so it points to where you saved BackupLoop1. Change the RUN WorkFlow in BackupLoop1 to point to BackupLoop2. Change the Run Workflow in BackupStart to point to BackupLoop1 and place BackupStart in your DOCK. When you get ready to record, drop the session file onto your dock icon, pick your backup folder(s), and then let it go. Hit record, and when you hit stop, wait a few seconds for it to end the raw file tags. Now check your backup folder. You should have a perfect WAV capture there.
IF audition crashes, you can drag the files in your backup folder to the original place and continue.
Bring the clip into the edit side and highlight a small section of JUST NOISE then go to EFFECTS > NOISE REDUCTION > NOISE REDUCTION > CAPTURE PROFILE > OK ... Now highlight the whole clip and go back to the noise reduction filter, use the PREVEIW and BYPASS to check your settings and then OK ... The worst thing about the noise is like a dripping faucet at night you will be hyper sensitive to it but use extreme CAUTION to not over do it ... Start at about 60% and work up from there clicking the BYPASS off and on till you get enough of the noise out so it's not distracting but doesn't affect the voice if you take too much the voice can get really ugly ... Background is the worst since it covers a wide range of frequencies and trying to remove it will cause you to take away from something you may not want to take away from making it sound less than, or just plain ugly ... Going into spectral view might help also ... Anyway you look at it the background noise is going to be ....... WILLIE
As Willie pointed out, noise reduction is a tricky business
when the noise you want to remove is so overt. Linked is
an example of your clip using Audition's regular noise
reduction tools at (almost) max strength. You'll notice
that the offending ambience has been removed but what's
left in its hole is noise not dissimilar to mp3 compression.
- The mp3 compression-type noise existed before the clip
was saved as an mp3.
- I've left a couple of seconds of the original file at the
beginning for comparative reference.