| Author |
Topic  |
Iscander
Posts: 4
|
Posted - Tue Jun 11, 2002 2:38 am |
|
|
Hi, All!
How do you think, what is the best noise reduction
software?
There are lots of products of this kind.
For example, plugin from SonicFoundry,
plugin from DiGICleen, from DigiDesign and so on.
And what is the best one, especially for very noised recordings?
|
|
ozpeter
Location: Australia
Posts: 3200
|
Posted - Tue Jun 11, 2002 6:09 am |
|
|
|
I get the impression that there are some heavyweight sound restoration people on this forum, and the reason is that Cool Edit is their tool of choice for this purpose - no plugin required! If you want some other inexpensive tools in the bag, especially for relatively quick-and-dirty work, the ClickFix plug in for Cool Edit does work with CEP2 too, and I sometime use a stand-alone program called Wave Corrector, but Cool Edit always remains at least part of the process for me.
|
|
Graeme
Member
Location: Spain
Posts: 4663
|
Posted - Tue Jun 11, 2002 11:35 am |
|
|
| Quote: | | I get the impression that there are some heavyweight sound restoration people on this forum, and the reason is that Cool Edit is their tool of choice for this purpose - no plugin required! |
Speaking for myself, I'd say this wasn't the case. I would certainly recommend CEP1.2a to anyone, if it was going to be their only investment, but other software/plug-ins can deal with specific problems in a more appropriate way and anyone deeply involved in restoration would be very unwise not to keep an eye on the competition.
I use CEP as a prime tool, but certainly not to the total exclusion of other software or plug-ins. CEP's prime advantage, for me, is its excellent editing capabilities and for virtually anything I do, then CEP gets involved somewhere along the line, even if it's just for an editing exercise.
Just for the record - apart from software which is currently 'under evaluation' and some stuff which I simply can't afford - I run 12 restoration apps and over 100 plug-ins (admittedly, not all these are restoration oriented, but I think you get the picture). On average, I probably use two or three different apps for each project, based on the problems I find.
Yes, I have my favourites - and CEP is certainly among them - but the true answer to the question is that the problem dictates the solution and CEP may not be the best one.
As I started out saying, CEP (certainly in it's 1.2a guise) is probably the best of the affordable apps around for this sort of work. If you are going to buy just the one app, then this is the one to go for. With practice and dilgence, it will cover 80% of most peoples requirements. My personal experience of v2.0 has been less happy and I would only recommend it with a number of reservations as to its stability. In my opinion, it is a great pity that Syntrillium saw fit to drop the earlier version when they introduced the current one - although I can see why they did this.
OTOH, if you have a lot of money burning a hole in your pocket (or an extremely friendly bank manager) then CEDAR is unarguably the bees knees of the restoration world.
Few of us can run to this sort of thing though and it's really best to acquire an arsenal of software from which one can select the most suitable for the specific project.
|
|
Iscander
Posts: 4
|
Posted - Thu Jun 20, 2002 5:28 am |
|
|
Thanx a lot for your answers!
I've tried CEP 1.2a Noise Reduction,
Sonic Foundry NoiseDX v2,
Sound Laundry 2.0 De-Noiser and
ActiveX plug-in from DIGiCleen
for noise reduction.
A few words about my problem.
I recorded a talk of two persons and there is
heavy sound of waterfall in the background.
The speech is hardly understandable.
My aim is to make speech more understandable.
I don't need to eliminate waterfall noise at all.
So, the results.
Plug-in from DIGiCleen inserts too many artifacts
into record. As I understand, it doesn't analize
spectrum of noise.
Other three tools produced different outputs,
but nevertheless unsatisfactory.
I can't definitly say, which tool is the best one.
Most likely, Sonic Foundry noise reduction is the best.
In addition to basic controls, the same as in CEP,
is has some advanced.
Playing with these advanced controls I slightly
improved sheech quality.
But still I haven't found an appropriate tool
to solve my problem!
What else should I do?
|
|
HanzZ
Location: Netherlands
Posts: 334
|
Posted - Sat Jun 22, 2002 1:54 am |
|
|
So there's a waterfall and some hardly understandable speech? Best way to solve the problem: do the recording again :)
But that's probably no option, is it?
Okay, well, there's really not much you can do. Noise reduction is certainly not going to help you in any way. Try eq-ing to boost the speech and to reduce the waterfall - although even that seems like an impossible task to me.
Good luck,
-- HanzZ
_________________ I make music like apples fall from trees...
|
|
|
|
Iscander
Posts: 4
|
Posted - Wed Jun 26, 2002 8:38 am |
|
|
Yes, rerecording is impossible.
I beleive noise reduction is some kind of eq-ing.
It analyzes noise frequencys and filters them out, doesn't it?
I'll try to equalize sound, but which frequencys human
speech consists of?
|
|
post78
Location: USA
Posts: 2887
|
Posted - Wed Jun 26, 2002 1:48 pm |
|
|
If you can afford it you should check out the X-restoration series from Waves. I've never heard it in action, but they make amazing plugins.
_________________ Answer = 1. Probably.
|
|
|
|
Graeme
Member
Location: Spain
Posts: 4663
|
Posted - Fri Jun 28, 2002 12:24 pm |
|
|
| Quote: | I beleive noise reduction is some kind of eq-ing.
It analyzes noise frequencys and filters them out, doesn't it? |
You're, sort-of, on the right track. Most post-production noise reduction systems rely on sampling a bit of the 'noise' and then adding it, out of phase, to the original. This will work fine for something relatively constant (say, tape hiss) but is not so successful with something which is not constant. Your waterfall falls into the latter category and that is why you are not having much success with this technique.
EQ is probably the only way which might produce an acceptable solution, although it's hard to say without actually hearing the original. From the description given, it doesn't sound too hopeful.
| Quote: | I'll try to equalize sound, but which frequencys human
speech consists of? |
You are more interested in knocking down the water than boosting the speech and, in general, it is better to cut frequencies than to boost them anyway. I suggest you try a parametric EQ. Set it to boost , with a low Q (2-5 would be a good starting point) and then sweep the frequency and listen for a major increase in the waterfall sound. Then change it to cut the level at this frequency and see how the speech sounds against the water. You'll certainly have to experiment a bit before you hit anything useful.
|
|
SteveG
Location: United Kingdom
Posts: 6695
|
Posted - Fri Jun 28, 2002 2:42 pm |
|
|
Unfortunately, one of the problems with waterfalls is that they tend to have a lot of acoustic energy in the same band as speech, so there will be a very distinct limit to what you can achieve just by filtering. And since their noise spectrum is quite broad, you really don't want any sort of a steep roll-off on the filters, as the effects of this (at the turnover points) would be rather more irritating than the waterfall sound in the first place.
I think that if you can do anything at all with this, it will be more likely to have come from a moderate amount of NR, leaving enough waterfall to mask the removal artefacts, but I wouldn't hold out a lot of hope. Why NR? Because the speech is not correlated, but the waterfall relatively is, which means that a short sample will remove a significant amount of the correlated sound without affecting the uncorrelated speech anything like as much. But whatever you do, this is never going to be good, and you will get artefacts.
Steve
_________________
|
|
|
|
post78
Location: USA
Posts: 2887
|
Posted - Fri Jun 28, 2002 8:32 pm |
|
|
| Quote: | | Quote: | I beleive noise reduction is some kind of eq-ing.
It analyzes noise frequencys and filters them out, doesn't it? |
You're, sort-of, on the right track. Most post-production noise reduction systems rely on sampling a bit of the 'noise' and then adding it, out of phase, to the original. |
Cool Edit's noise reduction is more of a frequency gate isn't it?
_________________ Answer = 1. Probably.
|
|
|
|
post78
Location: USA
Posts: 2887
|
Posted - Mon Jul 01, 2002 2:29 am |
|
|
_________________ Answer = 1. Probably.
|
|
|
|
| |
Topic
|