Removing hissing sounsd
Hi all
Just bought n-track as I am remixing an language course for my dad. Basically, I am replacing Portuguese from Portugal with Portuguese from Brazil, keeping the original English voices.
The bits I’m inserting have a slight hissing sound (specially noticable right before speaker speaks and right after), as they were recorded by me (although in pretty decent conditions…accoustic foam, very silent surroundings, etc) and the original track was done professionally. Needless to say the professional track has hardly any hiss at all (just the tiniest little bit - but that’s to be expected, I guess). So when I mix the two you can just about make out that the Brazilian Portuguese bits were “artificially inserted” if you understand what I mean.
My question is, which effects/method should I use to minimize these noticible inserts? I’m sure it can be done!
Oh and my dad is the author of this language course, so there’s no funny business taking place here, just in case some of you were wondering!!! 
Many thanks guys
…
| Quote (iangoes @ Feb. 02 2005,16:57) |
| ... The bits I'm inserting have a *slight* hissing sound (specially noticable right before speaker speaks and right after), as they were recorded by me (although in pretty decent conditions...accoustic foam, very silent surroundings, etc) and the original track was done professionally. Needless to say the professional track has hardly any hiss at all (just the tiniest little bit - but that's to be expected, I guess). So when I mix the two you can just about make out that the Brazilian Portuguese bits were "artificially inserted" if you understand what I mean. |
Please tell us more about how you're recording it: the room, the mic/preamp/soundcard, where's the computer, any other noise sources in the room.
Even some inexpensive soundcards have close to a professional noise level. Without knowing more, I'm guessing that the noise is coming from the mic + preamp, and maybe the PC or other room noise.
It's far easier, and better to record a better signal than it is to clean it up afterward.
As far as denoising tools, I'm used to the great noise tools in CoolEdit96 and CoolEdit2000, but they're now Adobe Audition. The free Audacity has a noise removal tool which some have complemented (I haven't tried it yet). If you need to do alot of cleanup, Pristine Sounds 2005 has really great tools.
Go with the last post that you should try to achieve the best possible recording etc.
Anyway, I have used Audacity, but ran into problems with it noticably changing what I wanted to keep (distorted it).
Another method, perhaps, is to use envelopes to control volume levels when there is speech and when there is not. As you say, hiss is most noticiable just before and just after the speech. It’s a pain to get the envelopes right, as you can literally cut your audio short of the mark.
Good luck
I think you’d be very happy with the results using either a noise gate or “de-noising” by filtering. This is just a voice recording, not a musical masterpiece. The intent is to convey information without distraction, and the change in noise level would be distracting. The voice recording has to be good enough to clearly hear the pronunciation & articulation, but it doesn’t have to be a work of art.
I would rather not add hiss, but that’s an option, and you can do it in n-Track by mixing down using dithering. If you want to try this, try dither values in powers of 2 (e.g., 4, 8, 16, 32 …) until you find one that sounds right.
Of course, it does make sense to make sure there’s nothing wrong and easily correctable with your recording setup, as mentioned above. But if you think the hiss is pretty mild and only annoying against the silent background of the rest, then you probably don’t need to fix anything.
You could tell us what the output meters show during the hiss. Also tell us what the peak values you get while speaking are. This will give us an idea whether your noise levels are reasonable.
You probably want to do a little compression on your vocals. This will even out the volume of your “performance”, and will also make some of the quieter sounds in your speech more intelligible. A very good tool to do this and silence the noise between parts is the n-Track compressor – the one with the graph. Not the multiband compressor. (I’m not sure what it’s called in n-Track V4.) With this compressor, there’s a preset called “soft kneed compression with noise gate” or something like that. Try it, and make sure the “compensate gain” button is on. I bet that’ll do the trick for you, very simply. After turning this on, you’ll have to readjust the track fader so that your voice matches the English in volume.
HTH,
Jeff
I just wanted to thank you guys for your advice. Jeff’s compression advice hit the spot;after applying “soft knee” (last week I felt more like applying a “hard knee” on something) the difference between the two bits isn’t very noticable anymore!
Now I am going to play around with some of the other techniques adviced in here, namely; gating, filtering, envelopes, etc.
Cool stuff! 
regards
Ian