audicity to record streaming audio

[Dave]

I've been listening to internet radio via my PC lately, and wanted to try to
record some of it to put on my mp3 player for while I'm driving. I've
previously used a program called SDK downloader (a fantastic program, BTW,
and free to boot) to grab MMS audio streams... podcasts, concerts on demand,
that sort of thing. But I either couldn't figure out how to set it up to
grab real-time (live) streaming audio in WMA format from internet radio or
it's just not designed to do it.

No problem, I says to myself, I've got audacity installed (curiosity, never
really used it for anything particularly useful). I'll just record using
that. So far so good. I run audacity and set it up to record the stream
from my soundcard which works good but, jumpin' Jesus, 88Mb/minute of
recording. Well, I don't happen to have a terabyte of free space. The
default settings for recording in audacity are 192000 at 32-bit float bit
depth. The streams I'm recording are marginal fidelity, some are 128K (the
highest) but a lot of them are garbage... 32k and even lower. So I don't
want to lose any quality. I've scoured web pages looking for this specific
scenario, but gave up after an hour or so of looking. Can anyone tell me
what rate and bit depth will give relatively lossless recording, hopefully
allowing me to record at 10MB/min or less? The literature out there says
that 44.1KHz 16-bit yields CD quality, but I really don't know enough about
it to guess. 128kbps is marginal fidelity as far as I'm concerned... I
certainly wouldn't BUY any music that's only encoded at 128kpbs, so I'm
curious as to how the streaming bit-rates relate to "CD quality" and the
like..


Sample question: If a CD is sampled at 44.1KHz, how can that be far far far
far better quality than a 64kpbs music stream from the internet. Where is
the information getting lost? Is it that the bit depth is different?

Thanks for any replies.

Dave

[Richard Crowley]

"Dave" wrote ...
> I've been listening to internet radio via my PC lately,
.....
Can anyone tell me
> what rate and bit depth will give relatively lossless recording, hopefully
> allowing me to record at 10MB/min or less? The literature out there says
> that 44.1KHz 16-bit yields CD quality, but I really don't know enough
> about it to guess. 128kbps is marginal fidelity as far as I'm
> concerned... I certainly wouldn't BUY any music that's only encoded at
> 128kpbs, so I'm curious as to how the streaming bit-rates relate to "CD
> quality" and the like..
>
> Sample question: If a CD is sampled at 44.1KHz, how can that be far far
> far far better quality than a 64kpbs music stream from the internet.
> Where is the information getting lost? Is it that the bit depth is
> different?

Certainly record at 16-bit depth. 24 and 32 are overkill for this
application, and 8 is inadequate for anything but telephone.

I would try sample rates of 22K or even 11K and see if the quality
vs. space tradeoff is acceptable for you. Nobody can make that
decision for you.

OTOH, I just use Total Recorder and record directly to MP3,
so I don't fool around with Audacity at all.

[David Nebenzahl]

On 5/1/2009 1:47 PM Dave spake thus:

> I've been listening to internet radio via my PC lately, and wanted to try to
> record some of it to put on my mp3 player for while I'm driving. I've
> previously used a program called SDK downloader (a fantastic program, BTW,
> and free to boot) to grab MMS audio streams... podcasts, concerts on demand,
> that sort of thing. But I either couldn't figure out how to set it up to
> grab real-time (live) streaming audio in WMA format from internet radio or
> it's just not designed to do it.
>
> No problem, I says to myself, I've got audacity installed (curiosity, never
> really used it for anything particularly useful). I'll just record using
> that. So far so good. I run audacity and set it up to record the stream
> from my soundcard which works good but, jumpin' Jesus, 88Mb/minute of
> recording. Well, I don't happen to have a terabyte of free space. The
> default settings for recording in audacity are 192000 at 32-bit float bit
> depth. The streams I'm recording are marginal fidelity, some are 128K (the
> highest) but a lot of them are garbage... 32k and even lower. So I don't
> want to lose any quality.

That last statement kinda doesn't make sense: you wouldn't want to lose
quality if you had any to begin with, but you're saying that what you're
recording is of "marginal fidelity" anyhow, so it's less of a concern.

If what you're recording is at 32K, then all you need is 32K. There's no
point in recording at a higher resolution than your source, right?
(Assuming you can determine what exactly the bitrate of the source is.)


--
Save the Planet
Kill Yourself

- motto of the Church of Euthanasia (http://www.churchofeuthanasia.org/)

[Dave]

"David Nebenzahl" > wrote in message
.com...
> On 5/1/2009 1:47 PM Dave spake thus:
>
>
> That last statement kinda doesn't make sense: you wouldn't want to lose
> quality if you had any to begin with, but you're saying that what you're
> recording is of "marginal fidelity" anyhow, so it's less of a concern.
>
> If what you're recording is at 32K, then all you need is 32K. There's no
> point in recording at a higher resolution than your source, right?
> (Assuming you can determine what exactly the bitrate of the source is.)
>
>
I guess my last statement of which you speak illustrates the depth of my
non-understanding. You have a source which is 64KHz or 64,000 samples per
second. It sounds pretty crappy compared to a commercial CD, which is
recorded at 44,100 samples per second.

Why is this?

I'm obviously missing something very very basic. I've heard it tossed about
that 192Kbps is about where you need to be with mp3 encoding rate for CD
quality, yet CD's are recorded at ~20% of that sample rate. ???
Regarding my last statement, I meant that I didn't want to make my 32kpbs
feed sound any worse.

[Richard Crowley]

"Dave" wrote ...
> I guess my last statement of which you speak illustrates the depth of my
> non-understanding. You have a source which is 64KHz or 64,000 samples per
> second. It sounds pretty crappy compared to a commercial CD, which is
> recorded at 44,100 samples per second.
>
> Why is this?
>
> I'm obviously missing something very very basic. I've heard it tossed
> about that 192Kbps is about where you need to be with mp3 encoding rate
> for CD quality, yet CD's are recorded at ~20% of that sample rate. ???
> Regarding my last statement, I meant that I didn't want to make my 32kpbs
> feed sound any worse.

Sample rate of 44,100 K (CD quality) = 1,411.2 Kbps !
(assuming 2-channel stereo @ 16-bit depth)

You seem to be confusing uncompressed sampling rate with
compressed bit rate.

[Dave]

"Richard Crowley" > wrote in message
...
> "Dave" wrote ...
>> I guess my last statement of which you speak illustrates the depth of my
>> non-understanding. You have a source which is 64KHz or 64,000 samples
>> per second. It sounds pretty crappy compared to a commercial CD, which
>> is recorded at 44,100 samples per second.
>>
>> Why is this?
>>
>> I'm obviously missing something very very basic. I've heard it tossed
>> about that 192Kbps is about where you need to be with mp3 encoding rate
>> for CD quality, yet CD's are recorded at ~20% of that sample rate. ???
>> Regarding my last statement, I meant that I didn't want to make my 32kpbs
>> feed sound any worse.
>
> Sample rate of 44,100 K (CD quality) = 1,411.2 Kbps !
> (assuming 2-channel stereo @ 16-bit depth)
>
> You seem to be confusing uncompressed sampling rate with
> compressed bit rate.
>
Ooh, dim bulb just went on. Each sample is then comprised of 16 bits? That
is the relationship I was looking for.

So, uh, taking my 32kpbs streaming rate as an example, I'd divide this by
the bit depth and divide again by the number of channels to determine the
sample rate... recording at 16-bit depth:

32,000 / (16 x 2) = 1000 or 1K sample rate to maintain equivalent quality.
Yes?

[Jim Gibson[_2_]]

In article <V0LKl.25254$Db2.21655@edtnps83>, Dave
> wrote:


> I'm obviously missing something very very basic. I've heard it tossed about
> that 192Kbps is about where you need to be with mp3 encoding rate for CD
> quality, yet CD's are recorded at ~20% of that sample rate. ???
> Regarding my last statement, I meant that I didn't want to make my 32kpbs
> feed sound any worse.

CDs are recorded at 44,100 16-bit samples per second per channel. So
that is 44100 x 16 x 2 = 1,411,200 bits/second, considerably more than
192Kbps. The difference is compression, which CDs do not have but MP3
does.

--
Jim Gibson

[Richard Crowley]

"Dave" wrote ...
> "Richard Crowley" wrote ...
>> You seem to be confusing uncompressed sampling rate with
>> compressed bit rate.
>>
> Ooh, dim bulb just went on. Each sample is then comprised of 16 bits?

Each sample of "CD quality" is comprised of 16 bits.
There is no general assumption for number of bits per sample.

> That is the relationship I was looking for.

You may be oversimplifying the relationship. This only applies
to *uncompressed* streams which are sampled at 16-bit.

> So, uh, taking my 32kpbs streaming rate as an example, I'd divide this by
> the bit depth and divide again by the number of channels to determine the
> sample rate... recording at 16-bit depth:
>
> 32,000 / (16 x 2) = 1000 or 1K sample rate to maintain equivalent
> quality.

No! This is important: There is *NO* simple correlation between
uncompressed *SAMPLE* RATE and compressed *BIT* RATE.

A 1K sample rate will produce nothing resembling music unless you
are an elephant or a whale with hearing limited to ~500Hz.
http://en.wikipedia.org/wiki/Nyquist_frequency

Compressed *BIT* RATE is influenced by not only sample rate, but
ALSO by sample DEPTH and *compression RATE*.

If you want to chose an uncompressed sample rate that will
preserve *your percieved quality* of an de-compressed stream,
ONLY a subjective experiment and evaluation with YOUR EARS
will suffice. There is no mathematical calculation for this.

[[email protected]]

On May 1, 4:47*pm, "Dave" > wrote:
> Sample question: *If a CD is sampled at 44.1KHz, how can
> that be far far far far better quality than a 64kpbs music
> stream from the internet.

Well, you're comparing apples to orchards.

CD quality is two 16 bit samples 44,100 times
per second. Let's see, that's

2 * 16 * 44,100

or 1,411,200 bits per second.

>*Where is the information getting lost? *

in going from 1,411,200 bits per second, or
1,411.2 kbits/sec to 64 kbits/second, you're
getting rid of all put about 4.5% of the original
data.

> Is it that the bit depth is different?

It's not a sinple matter of "bit depth." It's about
finding and eliminating redundancy, finding and
eliminating information which could well be
audibly irrelevant (under some criteria or another),
and then packaging the result efficiently.

[Geoff]

Dave wrote:
> I've been listening to internet radio via my PC lately, and wanted to
> try to record some of it to put on my mp3 player for while I'm
> driving. I've previously used a program called SDK downloader (a
> fantastic program, BTW, and free to boot) to grab MMS audio
> streams... podcasts, concerts on demand, that sort of thing. But I
> either couldn't figure out how to set it up to grab real-time (live)
> streaming audio in WMA format from internet radio or it's just not
> designed to do it.


Well, presumably not that 'fantastic' then ?

geoff

[Geoff]

Dave wrote:
> "David Nebenzahl" > wrote in message
> .com...
>> On 5/1/2009 1:47 PM Dave spake thus:
>>
>>
>> That last statement kinda doesn't make sense: you wouldn't want to
>> lose quality if you had any to begin with, but you're saying that
>> what you're recording is of "marginal fidelity" anyhow, so it's less
>> of a concern. If what you're recording is at 32K, then all you need is
>> 32K.
>> There's no point in recording at a higher resolution than your
>> source, right? (Assuming you can determine what exactly the bitrate
>> of the source is.)
> I guess my last statement of which you speak illustrates the depth of
> my non-understanding. You have a source which is 64KHz or 64,000
> samples per second. It sounds pretty crappy compared to a commercial
> CD, which is recorded at 44,100 samples per second.
>
> Why is this?

You are maybe confusing encoding bit-rate with CD bit-depth.

geoff

[Dave]

"geoff" > wrote in message
...
> Dave wrote:
>> I've been listening to internet radio via my PC lately, and wanted to
>> try to record some of it to put on my mp3 player for while I'm
>> driving. I've previously used a program called SDK downloader (a
>> fantastic program, BTW, and free to boot) to grab MMS audio
>> streams... podcasts, concerts on demand, that sort of thing. But I
>> either couldn't figure out how to set it up to grab real-time (live)
>> streaming audio in WMA format from internet radio or it's just not
>> designed to do it.
>
>
> Well, presumably not that 'fantastic' then ?
>
> geoff
>
One could say that. One could say using your logic, then, that a Ferrari is
not a fantastic vehicle because it cannot be used to plow snow.

SDK Downloader has many nice features, such as a VCR function which allows
you to tape on-line broadcasts without having to be sitting at your computer
to start and stop things.

[Dave]

"Richard Crowley" > wrote in message
...
> No! This is important: There is *NO* simple correlation between
> uncompressed *SAMPLE* RATE and compressed *BIT* RATE.
>
>
> If you want to chose an uncompressed sample rate that will
> preserve *your percieved quality* of an de-compressed stream,
> ONLY a subjective experiment and evaluation with YOUR EARS
> will suffice. There is no mathematical calculation for this.
>
>
Hmm... sounds like worst-case scenario... no simple answer or formula.
Maybe I'll look at Total Recorder if it can capture the stream in whatever
native format it's being played... for example if it's an mp3 stream then
I'd like to be able to capture that stream pre-decompression so I'm not
decompressing it, grabbing the analog output of my soundcard and digitizing
it, then having to re-compress it (with associated loss of quality) to play
it on my mp3 player. That just sounds ugly.

To my ears, the difference between 64kpbs and 128 kpbs is night and day.
The difference between 128kpbs and 192kpbs is discernible in an A/B
comparison, but not so obvious as the 64k/128k difference. Mainly I'm
listening to internet radio at work through my computer speakers, but at
home I think a 128kpbs stream sounds pretty crappy through my Magnepans vs.
CD source I am used to listening to.

[Don Pearce[_3_]]

On Mon, 04 May 2009 19:20:45 GMT, "Dave" > wrote:

>To my ears, the difference between 64kpbs and 128 kpbs is night and day.
>The difference between 128kpbs and 192kpbs is discernible in an A/B
>comparison, but not so obvious as the 64k/128k difference. Mainly I'm
>listening to internet radio at work through my computer speakers, but at
>home I think a 128kpbs stream sounds pretty crappy through my Magnepans vs.
>CD source I am used to listening to.

It sounds like you have the kind of hearing acuity one would expect to
encounter in this kind of group, and are able to appraise it honestly.

So, go with what sounds right to you, would be my call.

d

[Dave]

"Don Pearce" > wrote in message
news:4a0c4126.200029015@localhost...
> On Mon, 04 May 2009 19:20:45 GMT, "Dave" > wrote:
>
>>To my ears, the difference between 64kpbs and 128 kpbs is night and day.
>>The difference between 128kpbs and 192kpbs is discernible in an A/B
>>comparison, but not so obvious as the 64k/128k difference. Mainly I'm
>>listening to internet radio at work through my computer speakers, but at
>>home I think a 128kpbs stream sounds pretty crappy through my Magnepans
>>vs.
>>CD source I am used to listening to.
>
> It sounds like you have the kind of hearing acuity one would expect to
> encounter in this kind of group, and are able to appraise it honestly.
>
> So, go with what sounds right to you, would be my call.
>
> d
Yes, I'll record at a few different bitrates and see how it sounds. Thanks
for all of the insight, folks.

[Geoff]

Dave wrote:
..
>>
>>
>> Well, presumably not that 'fantastic' then ?
>>
>> geoff
>>
> One could say that. One could say using your logic, then, that a
> Ferrari is not a fantastic vehicle because it cannot be used to plow
> snow.
> SDK Downloader has many nice features, such as a VCR function which
> allows you to tape on-line broadcasts without having to be sitting at
> your computer to start and stop things.

Well evidently not quite.

More like having a Ferrari that won't turn right.

geoff