[Mxsmanic]

Is it better to increase the gain on my Zoom in situations where sound levels
are low, or should I leave the gain lower and then normalize the recording
later on? Which produces less noise in the end result? Or do they both amount
to the same thing?

[PStamler]

On Mar 23, 12:42*am, Mxsmanic wrote:
Is it better to increase the gain on my Zoom in situations where sound levels
are low, or should I leave the gain lower and then normalize the recording
later on? Which produces less noise in the end result? Or do they both amount
to the same thing?

Increase the gain.

Peace,
Paul

[Peter Larsen[_3_]]

Mxsmanic wrote:

Is it better to increase the gain on my Zoom in situations where
sound levels are low, or should I leave the gain lower and then
normalize the recording later on?

There is no single answer to this. Optimum is to set the gain right. If you
can't get near the "upper bit" with available input signal you should - in
my opinion - investigate whether the external pre you are using is proper
for the task.

Which produces less noise in the
end result? Or do they both amount to the same thing?

I don't know what Zoom device you have. I don't know what microphone(s) you
have. I don't know which external pre you are using. I don't know what you
are recording.

Kind regards

Peter Larsen

[Mxsmanic]

Peter Larsen writes:

There is no single answer to this. Optimum is to set the gain right. If you
can't get near the "upper bit" with available input signal you should - in
my opinion - investigate whether the external pre you are using is proper
for the task.

OK

I don't know what Zoom device you have. I don't know what microphone(s) you
have. I don't know which external pre you are using. I don't know what you
are recording.

H4n, with its built-in microphones. Nothing external. Mostly city noises, like
traffic, or sounds in a park or on a bridge, etc.

If there's no fixed answer to this, is there an experiment that I can do that
would allow me to find the answer? For example, with digital cameras, you can
take a picture of darkness (lens covered) and then pull the result into
Photoshop and force the pixel values to span their full scale, producing a
visible pattern of weak pixels, if any. I don't know if there's a way to
"record silence" with built-in microphones, though.

[Luxey]

On Friday, 23 March 2012 06:42:33 UTC+1, Mxsmanic wrote:
Is it better to increase the gain on my Zoom in situations where sound levels
are low, or should I leave the gain lower and then normalize the recording
later on? Which produces less noise in the end result? Or do they both amount
to the same thing?

Increase gain, so that highest peaks are as close to zero as possible, but not cliping. So you have to make voluntary sfety margin, as small as you think will be enough.
Also you can ride the gain during recording to keep general level steady, if applicable. Gain pot must not be crackling one, though.

You can go external mic preamp route. In that case you set gain as suited for the main part, then, beteween preamp and recorder, insert
1. gate/ expander to cut the noise off/ make sound above threshold proportionaly louder and
2. compressor/ limiter, to keep general level steady and take care of peaks.

Ha, do we have Pro user vs. pro equipment situation, or don't we?

[Mxsmanic]

Soundhaspriority writes:

The two steps are not equivalent. Each has a different purpose.

Which will produce less noise in the finished recording? Or will the amount of
noise in the finished product be the same either way?

It's mainly a concern for very quiet environments, such as inside a park. For
other street noise, the sound levels are so high that any noise from the
equipment is irrelevant.

[Mxsmanic]

Luxey writes:

Increase gain, so that highest peaks are as close to zero as possible,
but not cliping. So you have to make voluntary sfety margin, as small
as you think will be enough.

OK. So in a very quiet environment, it's better to crank the gain than to set
it lower and normalize later on.

Also you can ride the gain during recording to keep general level
steady, if applicable. Gain pot must not be crackling one, though.

The Zoom uses a small rocker button to adjust gain and unfortunately it makes
a slight clinking noise if you adjust it during recording, but sometimes I do
resort to that rather than start over.

You can go external mic preamp route. In that case you set gain as suited
for the main part, then, beteween preamp and recorder, insert
1. gate/ expander to cut the noise off/ make sound above threshold
proportionaly louder and 2. compressor/ limiter, to keep general level
steady and take care of peaks.

Currently I am using only built-in mics for this type of field recording, for
the sake of convenience (and I can't afford anything else, anyway).

Ha, do we have Pro user vs. pro equipment situation, or don't we?

I suspect that the recorder is much more professional than I am. It seems to
produce very nice recordings.

[Mike Rivers]

On 3/23/2012 5:29 AM, Mxsmanic wrote:

H4n, with its built-in microphones. Nothing external. Mostly city noises, like
traffic, or sounds in a park or on a bridge, etc.

Set the gain as best you can with the controls on the
recorder. While you will get a higher background electrical
noise level when you turn up the record gain on the
recorder, you'll get the best signal-to-noise ratio. If you
don't have a good enough s/n ratio, that means you don't
have enough signal and you need to get closer to the source.

You should recognize that the SPL generated by the things
you're wanting to record is really not very great unless
you're talking about a passing fire engine with the siren
blaring (which can often exceed 100 dB SPL) so you're simply
not going to be able to bring them too far above ambient
noise. It's what's there.

If there's no fixed answer to this, is there an experiment that I can do that
would allow me to find the answer?

Sure. Record some of the kinds of sound that you want to
capture with the gain set at maximum (or as high as you can
get it before you reach clipping) and then back the gain off
by 10 or 20 dB. Import the recordings into your computer
audio program, add gain to the sections where you lowered
the gain, then listen to the playback. Pick what you like
the most.

I don't know if there's a way to
"record silence" with built-in microphones, though.

Wrap it in a blanket and put it in a quiet room. It's not
perfect but it'll let you listen to how much the noise
increases when you turn up the gain. You can do a similar
experiment this way to recording what you're trying to
record but with essentially no signal. But you'll probably
need to add more gain to all of the test recordings of
"silence" in order to clearly hear the difference. For
example, if you record at maximum gain, 10 dB, and 20 dB
lower, you'll probably want to add 60 dB to the lowest gain
recording, 50 dB to the next lowest, and 40 dB to the
highest gain setting. That should give you something loud
enough to hear the differences.




--
"Today's production equipment is IT based and cannot be
operated without a passing knowledge of computing, although
it seems that it can be operated without a passing knowledge
of audio." - John Watkinson

http://mikeriversaudio.wordpress.com - useful and
interesting audio stuff

[Mike Rivers]

On 3/23/2012 6:21 AM, Mxsmanic wrote:

Which will produce less noise in the finished recording? Or will the amount of
noise in the finished product be the same either way?

Normalization is nothing more than adding gain. The thing is
that the process, rather than you , decides how much gain to
add in order to get to the level that you set as the target
(if it's not automatically full scale).

It's mainly a concern for very quiet environments, such as inside a park. For
other street noise, the sound levels are so high that any noise from the
equipment is irrelevant.

Either way, you'll be amplifying everything so the ratio
between the hummingbird flitting around the rose bush in the
park and the lawn mower on the other side of the street will
be unchanged. If you want the humming bird to speak up,
you'll need to get closer to the bird and further from the
lawn mower.



--
"Today's production equipment is IT based and cannot be
operated without a passing knowledge of computing, although
it seems that it can be operated without a passing knowledge
of audio." - John Watkinson

http://mikeriversaudio.wordpress.com - useful and
interesting audio stuff

[Arny Krueger[_4_]]

"Mxsmanic" wrote in message
...


Is it better to increase the gain on my Zoom in situations where sound
levels
are low, or should I leave the gain lower and then normalize the recording
later on?

Normalization is a very blunt stick. It works where indicated, but has broad
effects otherwise.

Plan B: Increase the level by adding thought-out number of dB worth of gain.

Which produces less noise in the end result? Or do they both amount
to the same thing?

Sounds levels low, covers an impossibly large range to give just one answer
to.

If your actual peak levels are 16 or more DB down, then increase the gain of
the Zoom while recording until they are more like 10 dB down.

Adjust the gain of your recordings to suit during production, prior to
distribution.

[Jason[_14_]]

On Fri, 23 Mar 2012 07:35:02 -0400 "Mike Rivers"
wrote in article

....


I don't know if there's a way to
"record silence" with built-in microphones, though.

Wrap it in a blanket and put it in a quiet room. It's not
perfect but it'll let you listen to how much the noise
increases when you turn up the gain.

I use an H4n a lot for recording chamber ensembles (but with external
mics). I did the wrap-it-a-blanket test when I bought the H4n (external
mics sharing the blanket). I even wrote down the results, which I can now
no longer find, but IIRC noise didn't go up much until gain was set close
to maximum, and even then the increase wasn't large. I haven't tried
substituting resistors for the mics; that might be interesting.

I tried something else. Audition has an Amplitude Statistics function. It
finds, among other things, the largest and smallest samples in the file
and calculates the dynamic range. For the recordings I typically make,
usually in venues that are not particularly quiet, it reports dynamic
range of 50-60dB. I set the recording level to peak around -20dB and have
never seen clipping except for the applause (limiting and low-cut filter
are off). The lowest-amplitude parts are still dominated by ambient room
noise. The wrapped-in-a-blanket test, with input gain set to where I
usually have it for live recording, yields dynamic range of about 60-
65dB. Now I'm going to try the same thing with the gain all the way up.

[Sean Conolly]

"Mxsmanic" wrote in message
...
Luxey writes:

Increase gain, so that highest peaks are as close to zero as possible,
but not cliping. So you have to make voluntary sfety margin, as small
as you think will be enough.

OK. So in a very quiet environment, it's better to crank the gain than to
set
it lower and normalize later on.

Be VERY careful about how close you get to zero - the meters are probably
not as accurate as you think.

Setting the gains a little too low can be adjusted in post. Setting them a
little high can not.

Sean

[Peter Larsen[_3_]]

Mxsmanic wrote:

Peter Larsen writes:

I don't know what Zoom device you have. I don't know what
microphone(s) you have. I don't know which external pre you are
using. I don't know what you are recording.

H4n, with its built-in microphones. Nothing external. Mostly city
noises, like traffic, or sounds in a park or on a bridge, etc.

For cityscapes I'd find a set and forget setting, ie. I'd boost in post as
long as things sound OK.
We have come a long way from the 12 and 13 bit recorders, I recall recording
alternative jazzical with my SV3800 and noting that I had a wordlenght
problem because the spoken, unamplified introductions sounded robotic. I
would even then weigh proper loudness scaling to be more important than
ultra-fi considerations.

A storyteller - on my suggestion - bought a H2 and went recording old folks
telling their life stories with the gain set in the niddle of its three
positions, ie. underrecorded so that it peaked -32 dB. Fixed the obvious
(based on FFT analysis) response dips that came from placing it on a table
and normalized and she - and the libraries that paid the recordings - was
happy, it good quite good.

If there's no fixed answer to this, is there an experiment that I can
do that would allow me to find the answer? For example, with digital
cameras, you can take a picture of darkness (lens covered) and then
pull the result into Photoshop and force the pixel values to span
their full scale, producing a visible pattern of weak pixels, if any.
I don't know if there's a way to "record silence" with built-in
microphones, though.

You have two factors to consider here. One being the wordlength you need,
Arny used to have some examples of how short a wordlength you could get away
with on his abx site as I recall it. The other being that analog electronics
tend to sound cleanest if put a reasonable amount of signal through them.

To render cityscapes for playback you quite possibly need to learn to use a
multiband compressor, the one from izotope that comes with Audition 2/3 is
to my liking, but you need to learn to use it. You can get it from izotope
if you're using another audio software package and there are no doubt
multiple other competing products. Remember: some of the time you have to
artificialize gravely for things to sound natural.

Summary: normalize may be the better strategy, possibly better still: put it
through a multiband compressor and normalize afterwards.

Kind regards

Peter Larsen

[Mike Rivers]

On 3/23/2012 9:52 PM, Jason wrote:

The wrapped-in-a-blanket test, with input gain set to where I
usually have it for live recording, yields dynamic range of about 60-
65dB.

That's about what you can expect from a good analog tape
deck without noise reduction. No wonder that recording
system worked so well for so many years.



--
"Today's production equipment is IT based and cannot be
operated without a passing knowledge of computing, although
it seems that it can be operated without a passing knowledge
of audio." - John Watkinson

http://mikeriversaudio.wordpress.com - useful and
interesting audio stuff

[Mxsmanic]

Peter Larsen writes:

The other being that analog electronics
tend to sound cleanest if put a reasonable amount of signal through them.

I'm not sure I understand this. If I put a high level of sound through a
recorder with the gain set to something intermediate, does that produce more
noise than a low level of sound with a high gain setting, or what?

From what I've read I've understood that the greater the gain in any type of
amplifier, the more noise and/or distortion it will introduce into the
amplified signal (more than a linear increase). Is that true, or do I have it
wrong?

It's not a huge issue for me, but I'd like to know how to set the gain on my
H4n in order to minimize noise, even though I suspect that its noise floor is
probably very low to begin with, for my purposes.

To render cityscapes for playback you quite possibly need to learn to use a
multiband compressor, the one from izotope that comes with Audition 2/3 is
to my liking, but you need to learn to use it. You can get it from izotope
if you're using another audio software package and there are no doubt
multiple other competing products. Remember: some of the time you have to
artificialize gravely for things to sound natural.

I'm still trying to figure out how to compression and when to know that it's
needed. I have no funds for payware plug-ins but there are some included
plug-ins with Sound Forge (I have Audacity and the OEM version of Sound Forge
that comes with Sony Vegas MS Platinum).

Summary: normalize may be the better strategy, possibly better still: put it
through a multiband compressor and normalize afterwards.

OK, I'll do some experiments with that.

Often the biggest problem I have with normalization is that there's something
in the recording that happens to go near 0 dB, and so the normalization
changes almost nothing.

[Mxsmanic]

Sean Conolly writes:

Be VERY careful about how close you get to zero - the meters are probably
not as accurate as you think.

Setting the gains a little too low can be adjusted in post. Setting them a
little high can not.

The sources I'm recording (traffic noises and other urban noise) tend to be
highly unpredictable, and often I end up recording with the meter hovering
around -12 dB or even -20 dB just to avoid clipping when the occasional car
horn honks or a motor scooter goes by.

One things that surprises me is that the sounds that I would expect to see
hitting 0 dB often do not, whereas the sounds that I would not expect to see
clipped are in fact being clipped. For example, car horns don't seem to be as
loud as they sound to my ears, but sometimes the rumble of a passing bus
actually reaches 0 dB even though it doesn't seem that loud.

[William Sommerwerck]

From what I've read I've understood that the greater the gain in any
type of amplifier, the more noise and/or distortion it will introduce
into the amplified signal (more than a linear increase). Is that true,
or do I have it wrong?

Broadly speaking, wrong, because you're saying "any". That's like saying
"any" car with a 200hp engine will necessarily have poorer gas mileage than
"any" car with a 150hp engine.

Given a specific design... If the amplifier's gain is set with feedback,
then higher gain will result in higher noise and distortion.

If an amplifier has fixed gain, and the "system" gain is set with a pot,
then the absolute noise level and distortion will not be affected by the
gain.

[PStamler]

On Mar 24, 8:22*am, Mxsmanic wrote:

One things that surprises me is that the sounds that I would expect to see
hitting 0 dB often do not, whereas the sounds that I would not expect to see
clipped are in fact being clipped. For example, car horns don't seem to be as
loud as they sound to my ears, but sometimes the rumble of a passing bus
actually reaches 0 dB even though it doesn't seem that loud.

Google "Fletcher-Munson" for an explanation.

Peace,
Paul

[Mike Rivers]

On 3/24/2012 9:18 AM, Mxsmanic wrote:

It's not a huge issue for me, but I'd like to know how to set the gain on my
H4n in order to minimize noise, even though I suspect that its noise floor is
probably very low to begin with, for my purposes.

Well, you really don't have much choice since there's only
one place to set record level, which appears to adjust the
gain or attenuation of the input stage. So the same old
analog principle applies - for the best signal-to-noise
ratio, get as much gain as you can from the earliest stage
amplifying the source. But the difference in signal-to-noise
ratio between having the digial record level hitting 0 dBFS
all the time (and hopefully not going over) and running it
10 dB lower is really insignificant in the practical sense.

Running at a little lower level keeps you safer from digital
clipping and maybe the converters are a little more linear a
bit down from the end of the scale. But running it so that
the level never exceeds -40 dBFS is probably going to make
the noise of the converters start to come up if you amplify
it after the fact.

Often the biggest problem I have with normalization is that there's something
in the recording that happens to go near 0 dB, and so the normalization
changes almost nothing.

If it's a brief sound that you can reduce the level of
without affecting the program material, then do that first,
and then normalize. If there's a car crash in the midst of a
bird song, you can just reduce the level of the car crash
and then bring up the level of the whole file by
normalizing. But if the song gets 10 dB louder in parts,
it's probably supposed to be that way. I never consider
normalization to be a "mastering process" but rather a
shortcut to bring up the overall level of a recording that
has a normal amount of dynamic range but just an overall low
level.

You can look at the waveform envelope and see where
something sticks up way above the eyeball average of the
material, listen to that, and decide if you can, or should,
knock it down, or if it needs to be there, and it's at the
right level relative to everything else.

This is all stuff that you can do with your digital editing
tools that's much more difficult to do with analog tools, so
it's easy to get tempted to spend way too much time with it.
But the results might be worth it.



--
"Today's production equipment is IT based and cannot be
operated without a passing knowledge of computing, although
it seems that it can be operated without a passing knowledge
of audio." - John Watkinson

http://mikeriversaudio.wordpress.com - useful and
interesting audio stuff

[Mike Rivers]

On 3/24/2012 9:22 AM, Mxsmanic wrote:

The sources I'm recording (traffic noises and other urban noise) tend to be
highly unpredictable, and often I end up recording with the meter hovering
around -12 dB or even -20 dB just to avoid clipping when the occasional car
horn honks or a motor scooter goes by.

There's nothing at all wrong with that. But consider this -
do you want the background noise, or is that car horn or
motor scooter really important and something that you want
to save? If it is, then you either need to allow for it, or
engage a limiter to sit on those unexpected sounds that are
much higher than your background level if it sounds OK. If
you don't care about those random loud sounds, let 'em clip
and just edit them out.

The TASCAM DR-40, and maybe you can do the same trick with
other recorders that can record 4 tracks, lets you set up a
second pair of tracks recording at a lower level than your
primary set. If something overloads on your primary tracks,
chances are it'll be OK on the tracks with lower gain, and
you can edit between the two sets of tracks. The Sony
PCM-D50's limiter actually works like this automatically. It
records a buffer 10 or 20 dB below the main track. When it
detects an overload, it automatically replaces the
overloaded section with the backup and then normalizes the
spliced-in segment so it will go to full scale. It's really
cool.

One things that surprises me is that the sounds that I would expect to see
hitting 0 dB often do not, whereas the sounds that I would not expect to see
clipped are in fact being clipped. For example, car horns don't seem to be as
loud as they sound to my ears, but sometimes the rumble of a passing bus
actually reaches 0 dB even though it doesn't seem that loud.

That's because a car horn is a designed sound. It's supposed
to be easy to hear, which means that it sounds louder than
it actually is, though I've measured passing fire engines at
100 dB SPL when their sirens are going. A bus has a lot of
low frequency energy that moves a lot of air, but your ear
is less sensitive than an A/D converter in that frequency
range.



--
"Today's production equipment is IT based and cannot be
operated without a passing knowledge of computing, although
it seems that it can be operated without a passing knowledge
of audio." - John Watkinson

http://mikeriversaudio.wordpress.com - useful and
interesting audio stuff

[Marc Wielage[_2_]]

On Thu, 22 Mar 2012 22:42:33 -0700, Mxsmanic wrote
(in article ):

Is it better to increase the gain on my Zoom in situations where sound levels
are low, or should I leave the gain lower and then normalize the recording
later on? Which produces less noise in the end result? Or do they both amount
to the same thing?
------------------------------snip------------------------------

See: gain staging.

http://www.emusician.com/news/0766/max-headroom/147106


--MFW

[Peter Larsen[_3_]]

Mxsmanic wrote:

Peter Larsen writes:

The other being that analog electronics
tend to sound cleanest if put a reasonable amount of signal through
them.

I'm not sure I understand this.

Read it aloud to yourself then, I mean exactly what I write. A couple of IC
opamp generations ago I used LF356's in my electronic x-over. When I moved
to this apartment I suddenly got neighbor complaints due to lower outside
background noise. To make it easier for myself to accomodate those I decided
to attenuate all power-amplifiers on the input and ended up with 15 dB
attenuation. Result: cleaner sound because the x-over ran 15 dB hotter.

If I put a high level of sound
through a recorder with the gain set to something intermediate, does
that produce more noise than a low level of sound with a high gain
setting, or what?

You ask how it measures when I say how it sounds.

From what I've read I've understood that the greater the gain in any
type of amplifier, the more noise and/or distortion it will introduce
into the amplified signal (more than a linear increase). Is that
true, or do I have it wrong?

I think you're closer to having it wrong than right in as much as it is just
not that simple. Remove the any as William suggested and it gets less wrong.
An example: The Fostex MR8HD is amazingly clean at its price point if you
feed it a line signal, but the higher the gain setting the grainier it gets.
It doesn't get "bad", just less clean, more "plasticky" sound on violins and
vox, just a bit. I have made recordings of storytelling on mine using it
near max gain, and used it for a vox improvisation that led to a great
"hook" yestersay totally maxed because I grabbed a SM7 and it sounds great.

It's not a huge issue for me, but I'd like to know how to set the
gain on my H4n in order to minimize noise, even though I suspect that
its noise floor is probably very low to begin with, for my purposes.

You set it so that the signal doesn't clip, except when/if as Mike suggested
you don't mind discarding clipped signal. When recording cityscapes - and
now it gets to be an artistic opinon rather than a techie opinion - I'd want
the fire engine and the diesel 18-wheeler recorded cleanly to preserve the
contrast. The result will be as unplayable as a fireworks recording or a
recording of avant garde jazzical, to get it playable you need to learn how
to scale it.

I think Mike's suggestion of recording a second pair of tracks 20 dB lower
is a good one. Parallel compression is another ploy, I think it is a british
invention, but it is so obvius once it is understood and so simple that many
people may have come up with it. Split the signal, put it through a
compresser with gentle settings and a low treshold and add it to the signal.
The outcome will be that transients are mostly uncompressed but the low
levels are brought up and overall the compression rate gets about half what
the compressor is set at. With Audition I have done it differently and just
drawn a suitable compressor transfer characteristic and never checked
whether truly modelling parallel compression would be better, perhaps with
the multiband.

I'm still trying to figure out how to compression and when to know
that it's needed. I have no funds for payware plug-ins but there are
some included plug-ins with Sound Forge (I have Audacity and the OEM
version of Sound Forge that comes with Sony Vegas MS Platinum).

There's a good three band in Magix "home studio" and the package ought to be
in the USD 50 to 79.95 range. It is however somewhat dumbified, and thus a
bit difficult to ensure that you're in 32 bit file format, I bought it some
years ago and ended up not at all using it.

Often the biggest problem I have with normalization is that there's
something in the recording that happens to go near 0 dB, and so the
normalization changes almost nothing.

That is not a problem, that is what it is. Anyway, it is, in the context of
making stuff sound similarly loud, useless anyway. Average level with a 300
millesecond time window as calculated by Auditions statistics window works
well for me. Auditions automated average level alignment for some reason
doesn't, especially not if allow to assume fletcher munson valid as a part
of the calculation, but also not anyway. Listening has been suggested as a
usable strategy, but nobody does that any more.

Kind regards

Peter Larsen

[Mxsmanic]

Mike Rivers writes:

If it's a brief sound that you can reduce the level of
without affecting the program material, then do that first,
and then normalize.

I haven't been able to figure out which function (in Sound Forge or Audacity)
does that. I can amplify a section of the recording, but there's no smoothing
of the edges or anything, so the entire recording suddenly becomes software
ore louder for a brief period, and this is easily audible and sounds
artificial.

Is there a standard function of some kind in sound-editing software that does
this a little more smoothly?

Often I resort to just deleting the loud sound. This seems to be less
noticeable than trying to change its intensity, even though it would sometimes
be nice to keep it.

I never consider
normalization to be a "mastering process" but rather a
shortcut to bring up the overall level of a recording that
has a normal amount of dynamic range but just an overall low
level.

For unpredictable sources I leave lots of headroom, but that compels me to
normalize later on if I want the recording to be at a "standard"
level--otherwise presumably someone else using it would have to normalize,
anyway.

Maybe I just need to find things to record that are more predictable. Believe
it or not, I haven't recorded any music with the H4n so far, although I have
recorded voice narration (with excellent results). Music raises the spectre of
copyright, which severely limits my ability to share anything that I record.

You can look at the waveform envelope and see where
something sticks up way above the eyeball average of the
material, listen to that, and decide if you can, or should,
knock it down, or if it needs to be there, and it's at the
right level relative to everything else.

But if it's something like, say, a pop or bang on the recording, I haven't
figured out how to crank down that single sound without reducing the audio
levels nearby.

This is all stuff that you can do with your digital editing
tools that's much more difficult to do with analog tools, so
it's easy to get tempted to spend way too much time with it.
But the results might be worth it.

Is there a specific tool I should look for that is particularly adapted to
reducing the intensity of loud, brief sounds in a recording?

[Mxsmanic]

Mike Rivers writes:

There's nothing at all wrong with that. But consider this -
do you want the background noise, or is that car horn or
motor scooter really important and something that you want
to save? If it is, then you either need to allow for it, or
engage a limiter to sit on those unexpected sounds that are
much higher than your background level if it sounds OK. If
you don't care about those random loud sounds, let 'em clip
and just edit them out.

Yes, I see your point.

The H4n has a built-in compressor and limiter, and some auto-level controls,
but I haven't used them thus far on the theory that I should not tinker with
the original recording any more than absolutely necessary, since changes made
during recording (as opposed to during editing) cannot be backed out. (This is
a philosophy I have followed with video for ages.)

The TASCAM DR-40, and maybe you can do the same trick with
other recorders that can record 4 tracks, lets you set up a
second pair of tracks recording at a lower level than your
primary set. If something overloads on your primary tracks,
chances are it'll be OK on the tracks with lower gain, and
you can edit between the two sets of tracks.

I can't find this on the H4n, unless I don't know what to look for. There's
some sort of adjustment for "mid/side" recording, but I haven't figured out
what it does.

The Sony
PCM-D50's limiter actually works like this automatically. It
records a buffer 10 or 20 dB below the main track. When it
detects an overload, it automatically replaces the
overloaded section with the backup and then normalizes the
spliced-in segment so it will go to full scale. It's really
cool.

Doesn't sound like I have that, but the H4n has all sorts of nifty functions
so I could be missing something.

That's because a car horn is a designed sound. It's supposed
to be easy to hear, which means that it sounds louder than
it actually is, though I've measured passing fire engines at
100 dB SPL when their sirens are going. A bus has a lot of
low frequency energy that moves a lot of air, but your ear
is less sensitive than an A/D converter in that frequency
range.

That makes sense.

I might be unusually sensitive to low frequencies because sometimes I hear
booming low frequencies that don't seem to bother anyone else, particularly
from vehicles. I guess their mufflers are designed to lower the frequency of
noise rather than eliminate it.

[Mxsmanic]

Marc Wielage writes:

See: gain staging.

http://www.emusician.com/news/0766/max-headroom/147106

Thanks!

[John Williamson]

Mxsmanic wrote:
Mike Rivers writes:

If it's a brief sound that you can reduce the level of
without affecting the program material, then do that first,
and then normalize.

I haven't been able to figure out which function (in Sound Forge or Audacity)
does that. I can amplify a section of the recording, but there's no smoothing
of the edges or anything, so the entire recording suddenly becomes software
ore louder for a brief period, and this is easily audible and sounds
artificial.

Is there a standard function of some kind in sound-editing software that does
this a little more smoothly?

Often I resort to just deleting the loud sound. This seems to be less
noticeable than trying to change its intensity, even though it would sometimes
be nice to keep it.

I never consider
normalization to be a "mastering process" but rather a
shortcut to bring up the overall level of a recording that
has a normal amount of dynamic range but just an overall low
level.

For unpredictable sources I leave lots of headroom, but that compels me to
normalize later on if I want the recording to be at a "standard"
level--otherwise presumably someone else using it would have to normalize,
anyway.

Maybe I just need to find things to record that are more predictable. Believe
it or not, I haven't recorded any music with the H4n so far, although I have
recorded voice narration (with excellent results). Music raises the spectre of
copyright, which severely limits my ability to share anything that I record.

You can look at the waveform envelope and see where
something sticks up way above the eyeball average of the
material, listen to that, and decide if you can, or should,
knock it down, or if it needs to be there, and it's at the
right level relative to everything else.

But if it's something like, say, a pop or bang on the recording, I haven't
figured out how to crank down that single sound without reducing the audio
levels nearby.

This is all stuff that you can do with your digital editing
tools that's much more difficult to do with analog tools, so
it's easy to get tempted to spend way too much time with it.
But the results might be worth it.

Is there a specific tool I should look for that is particularly adapted to
reducing the intensity of loud, brief sounds in a recording?

Volume automation lets you draw an envelope to control the channel gain.
It's available in all the audio editing programs I've ever used since
CoolEdit 96.

Or use a limiter plugin and turn the channel gain up, letting the
limiter automatically reduce the gain on the loud bits. When you do it
digitally, you can redo it as often as you like on copies using
different settings without generational losses or damaging the original
file.

--
Tciao for Now!

John.

[Mike Rivers]

On 3/25/2012 3:30 AM, Mxsmanic wrote:

If it's a brief sound that you can reduce the level of
without affecting the program material, then do that first,
and then normalize.

I haven't been able to figure out which function (in Sound Forge or Audacity)
does that.

I don't know about Audacity but in Sound Forge, just select
the part that's too loud by dragging the cursor across it
(it will be highlighted), then click on Process, then click
Volume in the Process menu, and adjust the slider. If you
don't highlight a region in the track, it will process the
whole recording. If you highlight a region, it will process
only the highlighted region.

I can amplify a section of the recording, but there's no smoothing
of the edges or anything, so the entire recording suddenly becomes software
ore louder for a brief period, and this is easily audible and sounds
artificial.

Well, it IS artificial. You can use Fade In and Fade Out to
make the transitions gentler, and zooming in on the area
where you want to work will help you find the best place to
make the changes. This is an editing process and it's a
craft. You need to practice.

Is there a standard function of some kind in sound-editing software that does
this a little more smoothly?

Yeah, it's called pay a professional if you don't want to
learn the techniques yourself.

For unpredictable sources I leave lots of headroom, but that compels me to
normalize later on if I want the recording to be at a "standard"
level--otherwise presumably someone else using it would have to normalize,
anyway.

There's nothing wrong with normalizing as long as you
understand what you're doing. Someone else using your
recording would in his own production would probably have to
adjust the level anyway.

Maybe I just need to find things to record that are more predictable.

That's the chicken way. Better to learn how to deal with
what you find in the field. You never know what you might
discover.

Is there a specific tool I should look for that is particularly adapted to
reducing the intensity of loud, brief sounds in a recording?

Volume adjustment or volume envelope. With a volume
envelope, you can draw a curve that represents the action of
a volume control, that that's really more useful when the
volume of an extended section (a couple of seconds or more)
or a single word needs to be adjusted.

--
"Today's production equipment is IT based and cannot be
operated without a passing knowledge of computing, although
it seems that it can be operated without a passing knowledge
of audio." - John Watkinson

http://mikeriversaudio.wordpress.com - useful and
interesting audio stuff

[Mike Rivers]

On 3/25/2012 3:44 AM, Mxsmanic wrote:

The H4n has a built-in compressor and limiter, and some auto-level controls,
but I haven't used them thus far on the theory that I should not tinker with
the original recording any more than absolutely necessary, since changes made
during recording (as opposed to during editing) cannot be backed out. (This is
a philosophy I have followed with video for ages.)

That's a good philosophy as far as it goes, but sometimes a
limiter or compressor can do what you would be doing
afterward anyway. Your real problem here as I see it is
simply lack of experience in doing this type of recording.
You're going to spoil some before you get the hang of it.
The more you do, the better your judgment will be.

One thing that's nearly always bad is an automatic volume
control, but a limiter or a compressor if set correctly can
be helpful. The problem is that when recording sounds like
you're doing, you really don't have an opportunity to adjust
the compressor based on the sound because it's only there
once and it's gone.

There's
some sort of adjustment for "mid/side" recording, but I haven't figured out
what it does.

That has to do with how the microphones "hear" stereo.

Doesn't sound like I have that, but the H4n has all sorts of nifty functions
so I could be missing something.

Read the manual and start experimenting to see what they do.


--
"Today's production equipment is IT based and cannot be
operated without a passing knowledge of computing, although
it seems that it can be operated without a passing knowledge
of audio." - John Watkinson

http://mikeriversaudio.wordpress.com - useful and
interesting audio stuff

[Luxey]

On Saturday, 24 March 2012 14:18:11 UTC+1, Mxsmanic wrote:
I'm not sure I understand this. If I put a high level of sound through a
recorder with the gain set to something intermediate, does that produce more
noise than a low level of sound with a high gain setting, or what?

More gain = more noise in absolute levels. However, proportionaaly, if you take a look at EIN numbers, you'll see the least noise is added at the highest gain.
To simplify, by adding Xdb of gain, you're adding less than Xdb of noise.
This is about the noise of preamp, not some background noise.

[Richard Webb[_3_]]

On Sun 2012-Mar-25 07:25, Mike Rivers writes:

On 3/25/2012 3:30 AM, Mxsmanic wrote:

If it's a brief sound that you can reduce the level of
without affecting the program material, then do that first,
and then normalize.

I haven't been able to figure out which function (in Sound Forge or Audacity)
does that.

I don't know about Audacity but in Sound Forge, just select the
part that's too loud by dragging the cursor across it
(it will be highlighted), then click on Process, then click Volume
in the Process menu, and adjust the slider. If you
don't highlight a region in the track, it will process the
whole recording. If you highlight a region, it will process only
the highlighted region.

Every daw I've ever been around has the ability to do
something similar, they may call it different things, but
that's where a daw has older analog methods beat hands down.

I can amplify a section of the recording, but there's no smoothing
of the edges or anything, so the entire recording suddenly becomes software
ore louder for a brief period, and this is easily audible and sounds
artificial.

Well, it IS artificial. You can use Fade In and Fade Out to make
the transitions gentler, and zooming in on the area
where you want to work will help you find the best place to make
the changes. This is an editing process and it's a
craft. You need to practice.

Yep, practice is how you learn, listen to your results.
Save the file under a different name if your daw doesn't
have 'undo" so that you can grab the unprocessed file and do it again after discarding the one you didn't like.
Adjusting volume envelopes adn the like is what sold the daw revolution.

Is there a standard function of some kind in sound-editing software that does
this a little more smoothly?

Yeah, it's called pay a professional if you don't want to
learn the techniques yourself.

That's right, but they're easy enough to learn. practice,
and practice some more. You'll get better over time with
it. DOn't like the changes? Don't save your work, do it
again. This is why it's call nondestructive editing.

There's nothing wrong with normalizing as long as you
understand what you're doing. Someone else using your
recording would in his own production would probably have to adjust
the level anyway.

This is true.

Maybe I just need to find things to record that are more predictable.

That's the chicken way. Better to learn how to deal with
what you find in the field. You never know what you might
discover.

Indeed, adn all the tools you need are at your disposal with most of today's audio editing software. YOu just need to
take some time to get familiar with the tools, how they work and what they can do for you. Undo is your friend. YOu
have tools at your disposal that the audio editor of half a
century ago could only dream of, it's a matter of learing to use them effectively, which can't be taught, or gleaned from reading newsgroup posts. Skills here are developed by using them, and listening to your results.
"Today's production equipment is IT based and cannot be
operated without a passing knowledge of computing, although it
seems that it can be operated without a passing knowledge of
audio." - John Watkinson

YEp, and here we have the proof of that statement. IN this
case though, the man might actually want to learn. IF so,
the only way to do it is to jump in and do it.


Regards,
Richard
--
| Remove .my.foot for email
| via Waldo's Place USA Fidonet-Internet Gateway Site
| Standard disclaimer: The views of this user are strictly his own.

[Arny Krueger[_4_]]

"Mxsmanic" wrote in message
...
Peter Larsen writes:

The other being that analog electronics
tend to sound cleanest if put a reasonable amount of signal through them.

I'm not sure I understand this. If I put a high level of sound through a
recorder with the gain set to something intermediate, does that produce
more
noise than a low level of sound with a high gain setting, or what?

Usually the only effect is that a fixed amount of noise that is generated
near the input is amplified by the gain of the amplifier. More gain, more
noise in the recording.

From what I've read I've understood that the greater the gain in any type
of
amplifier, the more noise and/or distortion it will introduce into the
amplified signal (more than a linear increase). Is that true, or do I have
it
wrong?

For a given amplifier that may be true. Obviously, higher gain amplifiers
may be biilt with lower distortion by using more advanced technology, or
more of it.

Consider an amplifier with an active element that provides a fixed amount of
gain. Gain is set by an attenuator at its input, before any active
circuitry. It's distortion is unaffected by the gain setting, but it is
affected by the size of the output signal.

The two strong sources of distortion in an amplifier are output voltage and
gain.


It's not a huge issue for me, but I'd like to know how to set the gain on
my
H4n in order to minimize noise, even though I suspect that its noise floor
is
probably very low to begin with, for my purposes.

Experiment. You can measure how much noise it makes by analyzing its
recordings. Record with loud signals, soft signals and no signal/ Record
with low, medium and high gain. Mix and match. What do you find?


To render cityscapes for playback you quite possibly need to learn to use
a
multiband compressor, the one from izotope that comes with Audition 2/3
is
to my liking, but you need to learn to use it. You can get it from
izotope
if you're using another audio software package and there are no doubt
multiple other competing products. Remember: some of the time you have to
artificialize gravely for things to sound natural.

I'm still trying to figure out how to compression and when to know that
it's
needed. I have no funds for payware plug-ins but there are some included
plug-ins with Sound Forge (I have Audacity and the OEM version of Sound
Forge
that comes with Sony Vegas MS Platinum).

There are ton of freeware VST plugins.



Often the biggest problem I have with normalization is that there's
something
in the recording that happens to go near 0 dB, and so the normalization
changes almost nothing.

Often the peaks are extremely short term (a few milliseconds) and lowering
just them will have minimal effects on the overall sound quality. My DAW r
has tools for lowering them. Does yours?

[Arny Krueger[_4_]]

"Mxsmanic" wrote in message
...
Sean Conolly writes:

Be VERY careful about how close you get to zero - the meters are probably
not as accurate as you think.

Setting the gains a little too low can be adjusted in post. Setting them
a
little high can not.

The sources I'm recording (traffic noises and other urban noise) tend to
be
highly unpredictable, and often I end up recording with the meter hovering
around -12 dB or even -20 dB just to avoid clipping when the occasional
car
horn honks or a motor scooter goes by.

Don't worry about that.

One things that surprises me is that the sounds that I would expect to see
hitting 0 dB often do not, whereas the sounds that I would not expect to
see
clipped are in fact being clipped. For example, car horns don't seem to be
as
loud as they sound to my ears,

That's because they are designed to be heard, despite minimal energy use.
Their tones are full of frequencies where your ear is most sensitive.

but sometimes the rumble of a passing bus
actually reaches 0 dB even though it doesn't seem that loud.

That's because your ears are relatively insensitive at low frequencies.

[Mxsmanic]

Arny Krueger writes:

That's because they are designed to be heard, despite minimal energy use.
Their tones are full of frequencies where your ear is most sensitive.

That makes sense. (Yes, I'm familiar with the Fletcher-Munson curves, it just
hadn't occurred to me.)

but sometimes the rumble of a passing bus
actually reaches 0 dB even though it doesn't seem that loud.

That's because your ears are relatively insensitive at low frequencies.

Which I suppose is the reason for mufflers shifting the frequency of exhaust
sounds.

I note that in my bathroom (which has a frosted window facing the street), I
sometimes hear booming low-frequency noise when there's a bus idling at the
traffic light outside (about 50 meters away). If I step outside the bathroom
(which is quite small), the booming mostly stops, but inside the bathroom it
is very loud. There must be some weird interaction between the acoustics of
the small bathroom and the exhaust noises of the bus. And it seems to be a
very low sound, so much so that it makes my ears feel "puffy" as it rumbles. I
haven't had a chance to rush into the bathroom with my recorder yet to see how
the sound looks objectively, but someday I will.

[John Williamson]

Mxsmanic wrote:
Arny Krueger writes:

That's because they are designed to be heard, despite minimal energy use.
Their tones are full of frequencies where your ear is most sensitive.

That makes sense. (Yes, I'm familiar with the Fletcher-Munson curves, it just
hadn't occurred to me.)

but sometimes the rumble of a passing bus
actually reaches 0 dB even though it doesn't seem that loud.
That's because your ears are relatively insensitive at low frequencies.

Which I suppose is the reason for mufflers shifting the frequency of exhaust
sounds.

They don't actually shift the frequencies, they just work as a low pass
filter. There is also sound made by some engine blocks when they flex
under the combustion chamber pressures, and some resonances in the
bodywork which are excited by the engine and transmission noise.

I note that in my bathroom (which has a frosted window facing the street), I
sometimes hear booming low-frequency noise when there's a bus idling at the
traffic light outside (about 50 meters away). If I step outside the bathroom
(which is quite small), the booming mostly stops, but inside the bathroom it
is very loud. There must be some weird interaction between the acoustics of
the small bathroom and the exhaust noises of the bus. And it seems to be a
very low sound, so much so that it makes my ears feel "puffy" as it rumbles. I
haven't had a chance to rush into the bathroom with my recorder yet to see how
the sound looks objectively, but someday I will.

Most bus and lorry engines idle at 650 or 700 RPM, which give a
fundamental exhaust note of about 33 - 35 Hz. If you enter your bathroom
dimensions into this:-

http://www.mcsquared.com/metricmodes.htm

You may well find a room mode at the second or third harmonic of the bus
engine frequency.

--
Tciao for Now!

John.

[Mxsmanic]

John Williamson writes:

Most bus and lorry engines idle at 650 or 700 RPM, which give a
fundamental exhaust note of about 33 - 35 Hz. If you enter your bathroom
dimensions into this:-

http://www.mcsquared.com/metricmodes.htm

You may well find a room mode at the second or third harmonic of the bus
engine frequency.

I plugged in the room dimensions (W 1.6 x L 1.8 x H 3.0 meters), but I'm not
sure what the output was telling me.

[John Williamson]

Mxsmanic wrote:
John Williamson writes:

Most bus and lorry engines idle at 650 or 700 RPM, which give a
fundamental exhaust note of about 33 - 35 Hz. If you enter your bathroom
dimensions into this:-

http://www.mcsquared.com/metricmodes.htm

You may well find a room mode at the second or third harmonic of the bus
engine frequency.

I plugged in the room dimensions (W 1.6 x L 1.8 x H 3.0 meters), but I'm not
sure what the output was telling me.

The lowest frequency mode is at 57.33 Hz, although you may get some
excitation at half that, which is a bit low for most bus engines.

--
Tciao for Now!

John.

[Don Pearce[_3_]]

On Tue, 27 Mar 2012 12:45:37 +0200, Mxsmanic
wrote:

John Williamson writes:

Most bus and lorry engines idle at 650 or 700 RPM, which give a
fundamental exhaust note of about 33 - 35 Hz. If you enter your bathroom
dimensions into this:-

http://www.mcsquared.com/metricmodes.htm

You may well find a room mode at the second or third harmonic of the bus
engine frequency.

I plugged in the room dimensions (W 1.6 x L 1.8 x H 3.0 meters), but I'm not
sure what the output was telling me.

The frequencies are those at which the effects of modes are greatest.
If you play a tone at one of the frequencies, you will find that the
volume changes enormously (from loud to inaudible in some cases) as
you walk around the room. The idea of room treatment is to iron out
those changes and leave the response as far as possible without the
peaks and dips. A really good room treatment might reduce them to 10dB
or so.

The bus note in the bathroom is going to be close to one of those
modal frequencies. Walk across or along the bathroom while it is
happening, and you should be able to make the sound come and go.

d

[Mxsmanic]

Don Pearce writes:

The bus note in the bathroom is going to be close to one of those
modal frequencies. Walk across or along the bathroom while it is
happening, and you should be able to make the sound come and go.

Hmm, I shall try that.