Sign in with
Sign up | Sign in
Your question

Why do some parts of streaming sound better

Last response: in Home Audio
Share
Anonymous
March 7, 2005 10:28:26 PM

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

I was listening to a news feed from NPR the other day over the internet
via Real Audio. I think that the bit rate was somewhere around 64kbs or
perhaps as low as 32kbs. Now, I understand that there will be some ...
no, make that a lot ... of quality issues at that rate, but I noticed
that the portion of the broadcast from the studio had more artifacts
than a soundbite that came from a speech someone gave. I noticed the
same thing occured later when part of a field interview aired. These
cuts sounded better than what came from the studio. The studio stuff
had the usual "swishyness" associated with low bit rate streams but the
other stuff didn't have nearly as much or just a little. I think I know
why but can somebody offer me an explanation?

Bill Brophy
Voice over talent/producer

More about : parts streaming sound

Anonymous
March 8, 2005 6:59:25 AM

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

"Announcer" <bbrophy@earthlink.net> wrote in message news:1110252506.518456.90170@f14g2000cwb.googlegroups.com...
> I was listening to a news feed from NPR the other day over the internet
> via Real Audio. I think that the bit rate was somewhere around 64kbs or
> perhaps as low as 32kbs. Now, I understand that there will be some ...
> no, make that a lot ... of quality issues at that rate, but I noticed
> that the portion of the broadcast from the studio had more artifacts
> than a soundbite that came from a speech someone gave. I noticed the
> same thing occured later when part of a field interview aired. These
> cuts sounded better than what came from the studio. The studio stuff
> had the usual "swishyness" associated with low bit rate streams but the
> other stuff didn't have nearly as much or just a little. I think I know
> why but can somebody offer me an explanation?
>
> Bill Brophy
> Voice over talent/producer
>

My guess is that it's the result of mono encoding. Mono at 32 isn't all
that bad on voice (witness the BBC internet news stream that's often
found at 24).

DM
Anonymous
March 8, 2005 8:59:21 AM

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

I think it was all mono because it was the NPR news feed.

Brophy
Related resources
Anonymous
March 8, 2005 11:21:09 AM

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

Bill, I'm no expert on Real Audio or MP3 or ATRAC, but I have seen the
source code for some other low-bandwidth audio compression schemes as
used in cell phone communications, and can tell you that it's like the
dog who walks on his hind legs: We shouldn't expect it to be graceful;
it's a wonder that it works at all.

Perhaps you remember in the early days of digital audio, one question
that cropped up fairly often on CompuServe and other discussion boards
was: How can a PCM system encode the tremendous sonic complexity of,
say, an entire orchestra? That question assumed that the digital system
was "taking notes" in some abstract way, e.g. "OK, now the first violin
is playing an A above middle C ..."

Of course CDs don't really work that way, and in fact the same question
could be asked about any ordinary microphone--how does a complex audio
signal fit into just a pair of wires, a single signal voltage at any
one moment?

When you get into these very low bit-rate data compression schemes you
find that it does, in a way, come a little closer to that old idea of
how digital audio might work. Instead of analyzing in terms of
instruments, though, it's in terms of noises, ranges of frequencies
(pitches), and time spans. In some cases the output is produced (and
the input is preserved) through very crude but effective
approximations: a prescribed tone burst and/or band-limited noise burst
is controlled by a prescribed volume envelope of some kind over some
prescribed number of milliseconds.

In those instances it really is a synthesized simulation of the input
rather than any kind of direct reproduction from stored audio sample
values. And the more you constrict the bit rate, the more you force the
system to veer away from any direct rendering of stored audio samples
and instead, to use cruder and cruder "schematic" approximations of the
original signal.

The techniques behind these approximations have been carefully
field-tested according to intelligibility and "pleasantness" criteria
(i.e. lack of extreme unpleasantness)--but no one ever claimed that at
32 kbps they can be at all sonically accurate. We experience the result
as meaningful signals "plus" various more or less bizarre-sounding
artifacts. But the more you narrow the bit rate, the more its entire
rendering of the original sound becomes an "artifact"! We forgive some
of its shortcomings because we get usable information and also because,
in most cases, we don't have the original to compare it with. But in
any audiophile sense of the term, little or nothing of the original
signal is preserved in the output--at the lowest bit rates, it is only
an algorithmic reconstruction of some of the original signal's
characteristics as "seen" by the encoder.

--best regards
Anonymous
March 8, 2005 12:12:07 PM

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

In article <1110252506.518456.90170@f14g2000cwb.googlegroups.com> bbrophy@earthlink.net writes:

> I was listening to a news feed from NPR the other day over the internet
> via Real Audio. I think that the bit rate was somewhere around 64kbs or
> perhaps as low as 32kbs. Now, I understand that there will be some ...
> no, make that a lot ... of quality issues at that rate, but I noticed
> that the portion of the broadcast from the studio had more artifacts
> than a soundbite that came from a speech someone gave. I noticed the
> same thing occured later when part of a field interview aired.

I've also wondered about something similar. I find that when listening
to a streaming radio station for a couple of hours, the level will
vary quite a bit, "artifacts" (like phasiness) will come and go, and
sometimes the stereo image will get a hole in the middle (like one
channel flipped polarity). I figure it's just something in the chain
adapting to available transfer rates.

It may also be that the Internet feed comes ahead of the final
limiting that makes it possible to get away with sloppy on-air
engineering. That would at least explain the difference in levels when
program material changes.


--
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
March 8, 2005 9:43:34 PM

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

Announcer wrote:
> I was listening to a news feed from NPR the other day over the internet
> via Real Audio. I think that the bit rate was somewhere around 64kbs or
> perhaps as low as 32kbs. Now, I understand that there will be some ...
> no, make that a lot ... of quality issues at that rate, but I noticed
> that the portion of the broadcast from the studio had more artifacts
> than a soundbite that came from a speech someone gave. I noticed the
> same thing occured later when part of a field interview aired. These
> cuts sounded better than what came from the studio. The studio stuff
> had the usual "swishyness" associated with low bit rate streams but the
> other stuff didn't have nearly as much or just a little. I think I know
> why but can somebody offer me an explanation?

Assuming they were using the same codec for both, one possibility is that the field audio got bandpass filtered before encoding and the studio stuff didn't. Especially relevant if the studio has a mic with a presence boost.
Anonymous
March 10, 2005 12:21:27 AM

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

Kurt ... You may be hitting on something here. My gut feeling was that
the field stuff (that had little ambient sound BTW) was not as full
audio spectrum as the studio stuff. This would be especially true if
the field recording had the highpass rumble filter enabled or if the
audio was EQed during the post production editing. The narration from
the studio was probably flat from high quality mics (although I don't
know what mics NPR in Washington uses). I remember reading somewhere
where NPR is careful not to use any processing in their live air
studios or on the mics in the production rooms in order to maintain
continuity of sound from studio to studio.

What effect would a bandpass filter have on reducing low bitrate
artifiacts?
!