I have a bunch of wav files of the 16/44.1 variety. Their response is flat
up to 20kHz. I would like to roll them off at 10kHz. Is there a simple way
to do this?

Thanks,

Norm Strong

[Arny Krueger]

wrote in message

I have a bunch of wav files of the 16/44.1 variety. Their response is flat
up to 20kHz. I would like to roll
them off at 10kHz. Is there a simple way to do this?

(1) Use editing software with appropriate filtering. I think even Audacity
has it.

(2) Downsample to 20 KHz SR, and then upsample

[Walt]

wrote:

I have a bunch of wav files of the 16/44.1 variety. Their response is flat
up to 20kHz. I would like to roll them off at 10kHz. Is there a simple way
to do this?

Yes, it's called a low pass filter. Any software editing program should
have a built in one.

The real question is why you want to do this.

//Walt

[[email protected]]

Walt wrote:
wrote:

I have a bunch of wav files of the 16/44.1 variety. Their response is flat
up to 20kHz. I would like to roll them off at 10kHz. Is there a simple way
to do this?

Some more information is needed. What you want is a filter,
in essence, but there are a number of questions you might
want to try to answer to make the job of finding the right filter
easier.

Do you want the result to be limited to a 10 kHz bandwidth but
still retain the 16 bit/44.1 kHa sample rate?

When you say "roll them off at 10 kHz," do you mean that you
need a brick-wall filter at 10 kHz, such that you need total or
near-total attenuation of everything above 10 kHz, or is -3 dB
at 10 kHz with a moderate roloff above there sufficient. Remember
that a 12 dB/octavce filter is only going to be down 12 dB at 20
20 kHz.

To cut to the chase, describe your filter requirements better, e.g.:

What is the minimum stop band ( 10 kHz) attenuation
needed?

How wide a transition band can you tolerate?

How much ripple, attenuation, etc., can you tolerate in
the pass band?

Is you primary or sole criteria removal of information
above 10 kHz, or are other properties of importance,
i.e., passband group delay, pre- and post-ringing?
phase response in the passband, etc.

And some other questions:

Is this something you need to use on a long term basis,
or do you just have a couple of examples that you need
to operate on? If the latter, posisbly some kind soul here
might volunteer to do it for you and send you the results.
Must it act in real time, i.e., on an active "live" audio stream
or can it act on existing files?

Now, many of these questions might be unimportant,
but you need better describe what you want.

Yes, it's called a low pass filter. Any software editing
program should have a built in one.

Well, maybe they SHOULD, not not every one DOES.
And whether what they DO have is suitable or not is a
different issue altorgether.

wrote in message
oups.com...

Walt wrote:
wrote:

I have a bunch of wav files of the 16/44.1 variety. Their response is
flat
up to 20kHz. I would like to roll them off at 10kHz. Is there a
simple way
to do this?

Some more information is needed. What you want is a filter,
in essence, but there are a number of questions you might
want to try to answer to make the job of finding the right filter
easier.

Do you want the result to be limited to a 10 kHz bandwidth but
still retain the 16 bit/44.1 kHa sample rate?

When you say "roll them off at 10 kHz," do you mean that you
need a brick-wall filter at 10 kHz, such that you need total or
near-total attenuation of everything above 10 kHz, or is -3 dB
at 10 kHz with a moderate roloff above there sufficient. Remember
that a 12 dB/octavce filter is only going to be down 12 dB at 20
20 kHz.

To cut to the chase, describe your filter requirements better, e.g.:

What is the minimum stop band ( 10 kHz) attenuation
needed?

How wide a transition band can you tolerate?

How much ripple, attenuation, etc., can you tolerate in
the pass band?

Is you primary or sole criteria removal of information
above 10 kHz, or are other properties of importance,
i.e., passband group delay, pre- and post-ringing?
phase response in the passband, etc.

And some other questions:

Is this something you need to use on a long term basis,
or do you just have a couple of examples that you need
to operate on? If the latter, posisbly some kind soul here
might volunteer to do it for you and send you the results.
Must it act in real time, i.e., on an active "live" audio stream
or can it act on existing files?

Now, many of these questions might be unimportant,
but you need better describe what you want.

Yes, it's called a low pass filter. Any software editing
program should have a built in one.

Well, maybe they SHOULD, not not every one DOES.
And whether what they DO have is suitable or not is a
different issue altorgether.

Here's the situation: Let's say you have an excellent digital recording of
a piano made with low noise equipment. You take this wav recording and
compress it using vbr mp3 at 16/44.1k. It compresses 20:1; the mp3 file is
one-twentieth the size of the original. (These are real life numbers.)

Now make an analog tape recording from the original wav file. The results
sound almost identical to the wav file; there is only a barely perceptible
tape hiss--so little that you wouldn't even notice it unless you were on the
lookout for it. When you redigitze this analog tape and run it through the
mp3 encoder it compresses only 10:1. This second vbr mp3 file is audibly
identical to the first one, but is twice the size of the first, solely
because of the encoder's attempt to digitize the tape hiss. This is what I
want to avoid.

From time to time, I will want to make a CDR from the compressed file, so I
imagine I will want to stick with 16/44.1k.

Does this explanation help?

Thanks a bunch,

Norm Strong

[bob]

wrote:

Here's the situation: Let's say you have an excellent digital recording of
a piano made with low noise equipment. You take this wav recording and
compress it using vbr mp3 at 16/44.1k.

Not sure what this means, but I don't have direct experience with
variable bitrate codecs.

It compresses 20:1; the mp3 file is
one-twentieth the size of the original. (These are real life numbers.)

That looks like a lot of compression, but ok.

Now make an analog tape recording from the original wav file. The results
sound almost identical to the wav file; there is only a barely perceptible
tape hiss--so little that you wouldn't even notice it unless you were on the
lookout for it. When you redigitze this analog tape and run it through the
mp3 encoder it compresses only 10:1. This second vbr mp3 file is audibly
identical to the first one,

Really? No tape hiss?

but is twice the size of the first, solely
because of the encoder's attempt to digitize the tape hiss.

"Solely" may be an overstatement. The only difference you may *hear*
between the original wav file and the analog tape is tape hiss, but
that's not the only difference. You've probably lost some high
frequency information (which your codec might wipe out, if it were
still there), and who knows what else. When you compress a different
file, you get a different result. So part of the reason the redigitized
file compresses less may be that it contains less of the sorts of
information that the codec you're using can compress.

This is what I
want to avoid.

Well, given what I've said above, I'm not sure you can avoid it
entirely. But if you want to get rid of tape hiss from a wav file
(which I presume is what you get when you redigitize the tape), most
audio recording software has some facility to do this. Often, you can
sample a period of silence (between tracks, say), and then "subtract" a
percentage of that from the file. Other programs have equalizers, or
allow you to suppress a particular frequency band. You may be able to
suppress white noise. Any of these tools should be used carefully
(i.e., don't try to suppress 100% of tape hiss).

What software are you using to record? Perhaps someone who uses it can
give you more specific advice.

From time to time, I will want to make a CDR from the compressed file, so I
imagine I will want to stick with 16/44.1k.

Again, not sure I follow this.

bob

[Steven Sullivan]

bob wrote:
wrote:

Here's the situation: Let's say you have an excellent digital recording of
a piano made with low noise equipment. You take this wav recording and
compress it using vbr mp3 at 16/44.1k.

Not sure what this means, but I don't have direct experience with
variable bitrate codecs.

It's not clear to me either, and I do.


It compresses 20:1; the mp3 file is
one-twentieth the size of the original. (These are real life numbers.)

That looks like a lot of compression, but ok.

Yep, that's a large amount of compression. VBR @ ~190kbps avg yields ~4.5:1
compression in my experience.


From time to time, I will want to make a CDR from the compressed file, so I
imagine I will want to stick with 16/44.1k.

Again, not sure I follow this.

Well, as I'm sure you know, if he wants to make a CDR of mp3 files, and have it play as an CD
rather than a 'data' disc (as required by most older CD/DVD players, which cannot play mp3s in
native form), he has to resample it to .wav at 16/44 before burning it to CDR (although
again,many of today's CD/DVD players can play discs full of raw mp3s too). I'm not sure what
he means by 'vbr 16/44.1' though; mp3s are specified by a constant or average bitrate, not
sampling rate/bitdepth. '16/44.1' is not an option. And more importantly, no mp3 codec or
setting I know of has a 'sampling rate' equivalent to 44.1 kHz (i.e, to yield a usable
bandwith of ~20 kHz). They all more or less eliminate commonly 'unheard' frequencies
(and do other cool masking-related stuff), usually above 19 kHz and quite often
lower than that.

(FWIW, LAME 3.97 at a vbr setting of ~190kbps avg, which has a range of 170-210 kbps,
generally sound fine to me, on those rare occasions I use lossy compression rather than
lossless)




___
-S
"As human beings, we understand the world through simile, analogy,
metaphor, narrative and, sometimes, claymation." - B. Mason

[Mr.T]

"Steven Sullivan" wrote in message
...
bob wrote:
wrote:

Here's the situation: Let's say you have an excellent digital
recording of
a piano made with low noise equipment. You take this wav recording
and
compress it using vbr mp3 at 16/44.1k.
It compresses 20:1; the mp3 file is
one-twentieth the size of the original. (These are real life
numbers.)

Yep, that's a large amount of compression. VBR @ ~190kbps avg yields
~4.5:1
compression in my experience.

20:1 is certainly a large amount of compression, but 190kbs is also a lot
more than 4.5:1 compression.
(1.4Mbs/0.19Mbs = 7.4:1 compression approximately)
Not sure where your "experience" enters into the equation?


I'm not sure what
he means by 'vbr 16/44.1' though; mp3s are specified by a constant or
average bitrate, not
sampling rate/bitdepth. '16/44.1' is not an option. And more importantly,
no mp3 codec or
setting I know of has a 'sampling rate' equivalent to 44.1 kHz (i.e, to
yield a usable
bandwith of ~20 kHz). They all more or less eliminate commonly 'unheard'
frequencies
(and do other cool masking-related stuff), usually above 19 kHz and quite
often
lower than that.

What a load of ********, an MP3 file does have a fixed sample rate, as well
as a fixed or variable bit rate. (most software MP3 players will even show
you what the sample rate is if you cared to look) If the player doesn't know
what it is, it can't possibly clock the reconstructed data at the correct
rate.

Obviously the OP was asking whether it is easier to convert a 44/16 MP3 file
back to 44/16 wave file, but in fact it's quite trivial to convert any
sample rate/bit depth/bit rate back to CD compatible file format, the only
decision you need to make is how much quality you are prepared to lose in
the original conversion to MP3, and why the hell you would accept that loss
on a CD.

MrT.