Sign in with
Sign up | Sign in
Your question

digital noise reduction question

Last response: in Home Audio
Share
Anonymous
April 1, 2005 7:50:05 AM

Archived from groups: rec.audio.pro (More info?)

Regarding digital NR (e.g., denoising plugins, CEDAR, NoNoise, etc) to reduce
hiss -- curious to know what experienced recording engineers/sound gurus think --

is an audible increase in the S/N via digital NR *always* accompanied by some unwanted ,
degradative artifacting affecting the music?

Or can digital NR ever be 'harmless' with respect to the music -- i.e.,
leaving the music sounding the same (or better), while appreciably reducing the hiss?

FWIW it has been my experience (using Coole Edit/Audition) that some light
passes of digital NR with Audition
can leave the music subjectively intact while reducing hiss. But i haven't ever
tested the difference rigorously.


--

-S
It's not my business to do intelligent work. -- D. Rumsfeld, testifying
before the House Armed Services Committee
Anonymous
April 1, 2005 7:50:06 AM

Archived from groups: rec.audio.pro (More info?)

"Steven Sullivan" <ssully@panix.com> wrote in message
news:D 2igdd$rfn$1@reader1.panix.com...
>
> Regarding digital NR (e.g., denoising plugins, CEDAR, NoNoise, etc) to
reduce
> hiss -- curious to know what experienced recording engineers/sound gurus
think --
>
> is an audible increase in the S/N via digital NR *always* accompanied by
some unwanted ,
> degradative artifacting affecting the music?
>
> Or can digital NR ever be 'harmless' with respect to the music -- i.e.,
> leaving the music sounding the same (or better), while appreciably
reducing the hiss?
>
> FWIW it has been my experience (using Coole Edit/Audition) that some light
> passes of digital NR with Audition
> can leave the music subjectively intact while reducing hiss. But i
haven't ever
> tested the difference rigorously.
>

The noise isn't actually removed, it's de-emphasized. The noise is still
incorporated into the music, it's just attenuated when there is no music at
each frequency band. It's not harmless, but often does more good than harm.
Anonymous
April 1, 2005 1:41:38 PM

Archived from groups: rec.audio.pro (More info?)

On Thu, 31 Mar 2005 22:50:05 -0500, Steven Sullivan wrote
(in article <d2igdd$rfn$1@reader1.panix.com>):

>
> Regarding digital NR (e.g., denoising plugins, CEDAR, NoNoise, etc) to
reduce
> hiss -- curious to know what experienced recording engineers/sound gurus
> think --
>
> is an audible increase in the S/N via digital NR *always* accompanied by some

> unwanted ,
> degradative artifacting affecting the music?
>
> Or can digital NR ever be 'harmless' with respect to the music -- i.e.,
> leaving the music sounding the same (or better), while appreciably reducing
> the hiss?
>
> FWIW it has been my experience (using Coole Edit/Audition) that some light
> passes of digital NR with Audition
> can leave the music subjectively intact while reducing hiss. But i haven't
> ever
> tested the difference rigorously.

Steve,

That's your next challenge. Get back to us with the results. :) 

Regards,

Ty Ford



-- Ty Ford's equipment reviews, audio samples, rates and other audiocentric
stuff are at www.tyford.com
Related resources
Anonymous
April 1, 2005 2:17:01 PM

Archived from groups: rec.audio.pro (More info?)

Steven Sullivan <ssully@panix.com> wrote:
>Regarding digital NR (e.g., denoising plugins, CEDAR, NoNoise, etc) to reduce
>hiss -- curious to know what experienced recording engineers/sound gurus think --
>
>is an audible increase in the S/N via digital NR *always* accompanied by some unwanted ,
>degradative artifacting affecting the music?

Yes, but the artifacts might or might not be audible. The whole notion here
is that you're making compromises and the compromises that you make need to
be carefully thought out and checked on good monitors.

>Or can digital NR ever be 'harmless' with respect to the music -- i.e.,
>leaving the music sounding the same (or better), while appreciably reducing the hiss?

It will never sound the same, but it will sound close enough for the job.
Things like click and pop removal are pretty innocuous. Broadband noise
reduction is anything but innocuous and decrackling can also do some terrible
damage if you aren't careful.

>FWIW it has been my experience (using Coole Edit/Audition) that some light
>passes of digital NR with Audition
>can leave the music subjectively intact while reducing hiss. But i haven't ever
>tested the difference rigorously.

Yes, and you may find several light passes give you better noise floors and
fewer artifacts than one heavy one, too. The key is to have good monitors
so you can hear what is really going on and you can back off when things get
out of hand.
--scott

--
"C'est un Nagra. C'est suisse, et tres, tres precis."
Anonymous
April 1, 2005 4:07:29 PM

Archived from groups: rec.audio.pro (More info?)

In article <d2igdd$rfn$1@reader1.panix.com> ssully@panix.com writes:

> is an audible increase in the S/N via digital NR *always* accompanied by some
> unwanted ,
> degradative artifacting affecting the music?

Yes, but if you're judicious about it, the cure is better than the
disease.

> Or can digital NR ever be 'harmless' with respect to the music -- i.e.,
> leaving the music sounding the same (or better), while appreciably reducing the
> hiss?

See above.

> FWIW it has been my experience (using Coole Edit/Audition) that some light
> passes of digital NR with Audition
> can leave the music subjectively intact while reducing hiss. But i haven't
> ever
> tested the difference rigorously.

Yes you have. If you can't hear it, it doesn't make audible artifacts.
But if you want to hear what it does, apply it more agressively and
then you'll know what to listen for.



--
I'm really Mike Rivers (mrivers@d-and-d.com)
However, until the spam goes away or Hell freezes over,
lots of IP addresses are blocked from this system. If
you e-mail me and it bounces, use your secret decoder ring
and reach me here: double-m-eleven-double-zero at yahoo
Anonymous
April 1, 2005 10:03:04 PM

Archived from groups: rec.audio.pro (More info?)

"Scott Dorsey" <kludge@panix.com> wrote in message
news:D 2jold$n8c$1@panix2.panix.com...

> >FWIW it has been my experience (using Coole Edit/Audition) that some
light
> >passes of digital NR with Audition
> >can leave the music subjectively intact while reducing hiss. But i
haven't ever
> >tested the difference rigorously.
>
> Yes, and you may find several light passes give you better noise floors
and
> fewer artifacts than one heavy one, too. The key is to have good monitors
> so you can hear what is really going on and you can back off when things
get
> out of hand.

In my experience, it's also heavily program-dependent. Some items let me do
up to 4dB of noise reduction at a pass, others gag on 1dB, and I haven't
noticed any pattern that would let me predict how a particular recording
will react.

Peace,
Paul
Anonymous
April 2, 2005 1:39:18 PM

Archived from groups: rec.audio.pro (More info?)

I have had one experience of using CEDAR noise reduction, along with
appropriate mastering, on the track "Drowsy Maggie" by Irish flute
player Matt Molloy. The "original" on his "Heathery Breeze" CD and
the version on "Wooden Flute Obsession" volume 1
(http://www.worldtrad.org) with hiss minimized/removed. Everyone's
been quite happy with the WFO version, and I'm personally quite pleased.

Kevin Krell
International Traditional Music Society, Inc.

Steven Sullivan wrote:
> Regarding digital NR (e.g., denoising plugins, CEDAR, NoNoise, etc) to reduce
> hiss -- curious to know what experienced recording engineers/sound gurus think --
>
> is an audible increase in the S/N via digital NR *always* accompanied by some unwanted ,
> degradative artifacting affecting the music?
>
> Or can digital NR ever be 'harmless' with respect to the music -- i.e.,
> leaving the music sounding the same (or better), while appreciably reducing the hiss?
>
> FWIW it has been my experience (using Coole Edit/Audition) that some light
> passes of digital NR with Audition
> can leave the music subjectively intact while reducing hiss. But i haven't ever
> tested the difference rigorously.
>
>
Anonymous
April 2, 2005 6:42:00 PM

Archived from groups: rec.audio.pro (More info?)

On Fri, 1 Apr 2005 03:50:05 +0000 (UTC), Steven Sullivan
<ssully@panix.com> wrote:

>
>Regarding digital NR (e.g., denoising plugins, CEDAR, NoNoise, etc) to reduce
>hiss -- curious to know what experienced recording engineers/sound gurus think --
>
>is an audible increase in the S/N via digital NR *always* accompanied by some unwanted ,
>degradative artifacting affecting the music?
>
>Or can digital NR ever be 'harmless' with respect to the music -- i.e.,
>leaving the music sounding the same (or better), while appreciably reducing the hiss?
>
>FWIW it has been my experience (using Coole Edit/Audition) that some light
>passes of digital NR with Audition
>can leave the music subjectively intact while reducing hiss. But i haven't ever
>tested the difference rigorously.

-- I think _every_ broadband noise reduction system exibits the comb
filtering effect. The artifacts can be concealed by good equalization
techniques but the sound does remain changed to a various extent; it
depends on how aggresive is one going in removing the noise. Multiple
pass / very light settings seem to have the least adverse effects. A
good monitoring and judgement is a must. Tests by direct a/b
comparison can reveal also slightest differences (often I do it in the
morning with very "fresh" ears and the result is also quite often
"Delete", grumbling "God Allmighty, was that really me the night
before?).

If you can't fight the noise, live with it. This is far better than
wrecked recordings. No money given into any such tools warrants
instant results.

Edi Zubovic, Crikvenica, Croatia
Anonymous
April 2, 2005 9:35:21 PM

Archived from groups: rec.audio.pro (More info?)

Question: In doing multiple light passes, would you use the original
noiseprint every time? Or take a new noiseprint from the noise-reduced file?

Peace,
Paul
Anonymous
April 2, 2005 9:51:58 PM

Archived from groups: rec.audio.pro (More info?)

"Paul Stamler" <pstamlerhell@pobox.com> wrote in message
news:tJA3e.504671$w62.136173@bgtnsc05-news.ops.worldnet.att.net...
> Question: In doing multiple light passes, would you use the original
> noiseprint every time? Or take a new noiseprint from the noise-reduced
> file?

New noise print.
Anonymous
April 3, 2005 3:57:15 AM

Archived from groups: rec.audio.pro (More info?)

On Sat, 02 Apr 2005 17:35:21 GMT, "Paul Stamler"
<pstamlerhell@pobox.com> wrote:

>Question: In doing multiple light passes, would you use the original
>noiseprint every time? Or take a new noiseprint from the noise-reduced file?
>
>Peace,
>Paul
>
-- Yes, I'd use the new one. But there are other possibilities,
especially at worn and noisy records, where the original noiseprint
could be temporarily saved and it could be experimented with, esp.
decreasing volume, inverting etc. There are many possibilities for
experimenting.

Then you'd replace the original noiseprint such altered one and try.
The utmost aim is to have the original sound preserved -- altered or
vanished quitetest parts, absence of recorded natural reverberation
and alike could be a warning. As I noticed the most common mistake is
made by trying to remove all the noise. But as the noise is very
broadband, the damage is obvious. To me, it is better to make the
noise unobtrusive enough so that it wouldn't distract a listener from
hearing and make it so that it would the least mask the sound picture.
As to impulse noise the risk is, if you remove "Ticks" merely by
lowpass filtering, you in many cases would have a "Tock".

And such filtering to less than 7 kHz alters the sound too much as it
gets unduly loud and aggressive. Low-Q filters are better choice in
such tasks. At many old recordings I have taped decades ago, I like
the sound of LP filters of the old EMT 930 turntables, which had them
built-in. It seems to me, that they can't be easily digitally
reproduced but they were very smooth indeed.

OK, impulse noises are easier to deal with, the big ones manually and
there are also good automatic decracklers such Algorithmix/Waves ones,
the Sonic Forge (now Sony) "Click and Crackle Removal" from their
noise reduction package sold separately and some others. Careless
usage of some declickers results in nasty increased and altered clicks
and distorsion and with decracklers, the result can sound worn and
dulled.

Interesting, in Sound Forge 8 there is the "Express FX Audio
Restoration" plugin, which sounds to me as if it is dealing with
formants too; but I don't like results -- perhaps it's only me.

Dealing with hiss is far more risky, though. To be honest, I avoid
removing it by use of any noise reduction tools, but I have some other
mehods to reduce it somewhat. However, so reduced noise means also a
dulled result too so a good balance is important.


Edi Zubovic, Crikvenica, Croatia
Anonymous
April 5, 2005 12:02:56 AM

Archived from groups: rec.audio.pro (More info?)

Scott Dorsey <kludge@panix.com> wrote:
> Steven Sullivan <ssully@panix.com> wrote:
> >Regarding digital NR (e.g., denoising plugins, CEDAR, NoNoise, etc) to reduce
> >hiss -- curious to know what experienced recording engineers/sound gurus think --
> >
> >is an audible increase in the S/N via digital NR *always* accompanied by some unwanted ,
> >degradative artifacting affecting the music?

> Yes, but the artifacts might or might not be audible. The whole notion here
> is that you're making compromises and the compromises that you make need to
> be carefully thought out and checked on good monitors.

> >Or can digital NR ever be 'harmless' with respect to the music -- i.e.,
> >leaving the music sounding the same (or better), while appreciably reducing the hiss?

> It will never sound the same, but it will sound close enough for the job.
> Things like click and pop removal are pretty innocuous. Broadband noise
> reduction is anything but innocuous and decrackling can also do some terrible
> damage if you aren't careful.

> >FWIW it has been my experience (using Coole Edit/Audition) that some light
> >passes of digital NR with Audition
> >can leave the music subjectively intact while reducing hiss. But i haven't ever
> >tested the difference rigorously.

> Yes, and you may find several light passes give you better noise floors and
> fewer artifacts than one heavy one, too. The key is to have good monitors
> so you can hear what is really going on and you can back off when things get
> out of hand.


I understand all that. But it's not quite what I'm asking. I'm asking what
'clsoe enough for the job' means.


Think of my question this way: Suppose I believe that digital broadband NR
*inevitably* degrades the music -- degrades the 'signal' as well as the noise.
Even when applied by careful, experienced ears using great monitors. On that basis
I advocate never using digital broadband NR. It's better to leave the hiss
as is. Always.

This isnt' my stance, btw. But it is pretty much the stance of many
audiophile/'purist' types -- including some who are pros, like Steve Hoffman.

Do other pros in their heart of hearts agree with that belief, even if they
still use digital NR? Are they accepting a compromise?
Or do they believe that you really *can* reduce hiss digitally without the
music taking a hit?


--

-S
It's not my business to do intelligent work. -- D. Rumsfeld, testifying
before the House Armed Services Committee
Anonymous
April 5, 2005 12:02:57 AM

Archived from groups: rec.audio.pro (More info?)

Steven Sullivan <ssully@panix.com> wrote:
>
>I understand all that. But it's not quite what I'm asking. I'm asking what
>'clsoe enough for the job' means.

That depends on what the job is. If you're doing something for a quick
broadcast job, close enough is a lot closer than it would be for a CD
issue.

My notion of 'close enough' is very different than what the NoNoise butchers
at Madacy Records consider close enough.

>Think of my question this way: Suppose I believe that digital broadband NR
>*inevitably* degrades the music -- degrades the 'signal' as well as the noise.
>Even when applied by careful, experienced ears using great monitors. On that basis
>I advocate never using digital broadband NR. It's better to leave the hiss
>as is. Always.

There are folks who feel this way, and that's fine. They can listen to the
originals. For a lot of material, I completely agree with them. Other
material may be in such bad condition that I disagree.

This is like asking how much chocolate is enough for chocolate ice cream.
It's a judgement call. And it's not the same for ice cream that goes in
an ice cream cake and ice cream that is eaten alone, too.

>This isnt' my stance, btw. But it is pretty much the stance of many
>audiophile/'purist' types -- including some who are pros, like Steve Hoffman.
>
>Do other pros in their heart of hearts agree with that belief, even if they
>still use digital NR? Are they accepting a compromise?
>Or do they believe that you really *can* reduce hiss digitally without the
>music taking a hit?

Everything you do, everywhere, everytime, is accepting a compromise. That
is part of the job. The difference between good and bad engineering is
knowing how to make the right compromises for the situation.
--scott

--
"C'est un Nagra. C'est suisse, et tres, tres precis."
Anonymous
April 5, 2005 12:50:25 AM

Archived from groups: rec.audio.pro (More info?)

In article <d2s6hg$sh5$1@reader1.panix.com> ssully@panix.com writes:

> I understand all that. But it's not quite what I'm asking. I'm asking what
> 'clsoe enough for the job' means.

Only you will know that when you get there.

> Think of my question this way: Suppose I believe that digital broadband NR
> *inevitably* degrades the music -- degrades the 'signal' as well as the noise.
>
> Even when applied by careful, experienced ears using great monitors. On that
> basis
> I advocate never using digital broadband NR. It's better to leave the hiss
> as is. Always.

Fine, if that's your choice. I've never felt compelled to try to
reduce analog tape hiss - it doesn't spoil the musical listening
experience for me. But there are others who feel that it should be
done if possible, and the end listener doesn't know to what extent, if
any, the program material has been degraded because he or she has
never heard the "before."

> Do other pros in their heart of hearts agree with that belief, even if they
> still use digital NR? Are they accepting a compromise?

Of course. But one of the things that professionals do is manage
tradeoffs and compromises. They make decisions as to what's best for
the target listener (who may not be a died-in-the-wood audiophile who
attended the recording session 40 years ago and remembers exactly what
it sounded like).


--
I'm really Mike Rivers (mrivers@d-and-d.com)
However, until the spam goes away or Hell freezes over,
lots of IP addresses are blocked from this system. If
you e-mail me and it bounces, use your secret decoder ring
and reach me here: double-m-eleven-double-zero at yahoo
Anonymous
April 6, 2005 8:55:56 AM

Archived from groups: rec.audio.pro (More info?)

On Apr 4, 2005, Steven Sullivan <ssully@panix.com> commented:

> Think of my question this way: Suppose I believe that digital broadband NR
> *inevitably* degrades the music -- degrades the 'signal' as well as the
> noise.
> Even when applied by careful, experienced ears using great monitors. On that
> basis
> I advocate never using digital broadband NR. It's better to leave the hiss
> as is. Always.
>--------------------------------snip----------------------------------<

I learned a long time ago that "never" and "always" are two words you have to
use very, very carefully -- especially with engineering and artistic
decisions.

What Scott says elsewhere is true: noise-reduction is like any tool. You
can use it well, or you can use it badly, like EQ or compression or anything
else. The great thing about all these tools is that they have knobs, and you
don't have to run them up to "11" all the time. Sometimes a little bit works
best; sometimes switching them in at some point, and bypassing them elsewhere
is best. I find there's rarely a "one size fits all" adjustment for an
entire song, especially with NR. But to say there's never a time when
noise-reduction can work well is not true.

I think the real key, though, is just getting good source tapes. Obviously,
a 2nd-generation tape, played back clean, is going to sound better than a
5th-generation tape that's been noise-reduced to death. I think a lot of
times, when you hear a muddy, heavily-gated song on a CD, the reason why that
happened is because the label wasn't willing to take more time to search for
a better source tape. Great source tapes rarely need any NR or drastic EQ,
and on that, I'll generally agree. (But note that I don't use the words
"never" or "always.")

--MFW
Anonymous
April 6, 2005 12:59:54 PM

Archived from groups: rec.audio.pro (More info?)

Marc Wielage <mfw@musictrax.com> wrote:
>
>I think the real key, though, is just getting good source tapes. Obviously,
>a 2nd-generation tape, played back clean, is going to sound better than a
>5th-generation tape that's been noise-reduced to death. I think a lot of
>times, when you hear a muddy, heavily-gated song on a CD, the reason why that
>happened is because the label wasn't willing to take more time to search for
>a better source tape. Great source tapes rarely need any NR or drastic EQ,
>and on that, I'll generally agree. (But note that I don't use the words
>"never" or "always.")

Noise reduction can reduce noise. But it can't do _anything_ about distortion.

I often have people sending me 78 rpm acetates that were transcribed (using
1930s technology) from a 16" 33 rpm transcription disc. Comparing the dub
with the original, it's amazing how much generation loss there was back then.
And often you can hear that the original is mistracking on playback, but there
is nothing you can do about it.

Another personal gripe is getting tapes that were dubs made with incorrect
azimuth settings.

A clean transcription from the earliest possible generation goes a long way
toward getting good sound. NR is just a nice addition now and then.
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."
Anonymous
April 18, 2005 1:04:18 AM

Archived from groups: rec.audio.pro (More info?)

On Apr 6, 2005, Scott Dorsey <kludge@panix.com> commented:

> Another personal gripe is getting tapes that were dubs made with incorrect
> azimuth settings.
>
> A clean transcription from the earliest possible generation goes a long way
> toward getting good sound. NR is just a nice addition now and then.
>--------------------------------snip----------------------------------<

Yeah, amen to that. As an illustration: about ten years ago, I was working
on a project for 20th Century-Fox, doing a digital restoration on the 1954
Richard Burton film PRINCE OF PLAYERS. They sent over a stereo track made in
the 1970s, and it really sounded horrible. After trying to use it for a day
or two, I called the studio and asked, "hey, is this all you have? Because
it really sounds terrible." Basically, I think they had done the transfer
and not checked the azimuth correctly, and the tape sounded muddy and dull.

Fox checked their vault and told me the only other thing they had in stereo
was the original 4-track 1954 mag master! I gulped and had them send over
that. I had to clean the mag heads every five minutes or show because the
mag was disintegrating slightly, but it sounded about 500% better than the
1970s transfer. In fact, for several weeks afterwards, I used this track as
a demo (on digital tape) to show people just how could a 40 year-old
recording could sound. Quite an illuminating experience.

--MFW
Anonymous
April 18, 2005 1:04:19 AM

Archived from groups: rec.audio.pro (More info?)

"Marc Wielage" wrote ...
> Fox checked their vault and told me the only other thing they
> had in stereo was the original 4-track 1954 mag master!

I particularly liked the use of the phrase "only other"! :-)
Anonymous
April 18, 2005 1:47:57 PM

Archived from groups: rec.audio.pro (More info?)

On Sun, 17 Apr 2005 21:04:18 GMT, Marc Wielage <mfw@musictrax.com>
wrote:

>On Apr 6, 2005, Scott Dorsey <kludge@panix.com> commented:
>
>> Another personal gripe is getting tapes that were dubs made with incorrect
>> azimuth settings.
>>
>> A clean transcription from the earliest possible generation goes a long way
>> toward getting good sound. NR is just a nice addition now and then.
>>--------------------------------snip----------------------------------<
-----------8<---------------------too-------------------------------------
>Fox checked their vault and told me the only other thing they had in stereo
>was the original 4-track 1954 mag master! I gulped and had them send over
>that. I had to clean the mag heads every five minutes or show because the
>mag was disintegrating slightly, but it sounded about 500% better than the
>1970s transfer. In fact, for several weeks afterwards, I used this track as
>a demo (on digital tape) to show people just how could a 40 year-old
>recording could sound. Quite an illuminating experience.
>
>--MFW
>
>
See? --You've got a bingo. Unfortunately, getting a decent (ie. early
generation) dub, not to speak about a master, isn't easy. Often, one
has to pull something out of nothing because "nothing" is all that's
available. I had to better out a gramophone record, mono, from that
times, where the cutting tape deck has been obviously misaligned and
cut that way -- a bad work. I did what I could but the sound is
f-l-o-a-t-i-n-g. I think that somewhere a original tape _might_ exist
(sometimes they don't any more), but then, the masters are of course
treated as an asset.

Edi Zubovic, crikvenica, Croatia
!