Mailing List Archive

Cellphones and Audio (was Ghost Click, though I got no idea why)
----- Original Message -----
> From: "Valdis Kletnieks" <Valdis.Kletnieks@vt.edu>

> On Wed, 02 May 2012 13:10:28 -0700, Jeroen van Aart said:
> > Technical specs aside I believe you are mistaken with regards to the
> > actual every day reality. My experience (and anyone else I talked to)
> > calling to and from mobile phones has been 100% a bad one with regards
> > to audio quality.
>
> I look at my Samsung cell phone, and the tiny speaker squeezed in up over the
> screen at one end, and then I think of the large speakers in the handset of an
> old-school Bell system rotary phone. Then I think about the fact that my
> laptop has pretty damned good sound quality when I plug in a good pair of
> Kenwood KPM-410 headphones, and sounds totally crappy over the tiny built-in
> speakers that Dell provided.
>
> It may not be the codec that sucks...

Right.

Me and my business partner have both spent quite a number of years involved
with sound reinforcement and other types of audio engineering, and we're
therefore better positioned to evaluate the transmit and receive audio of
various communications channels and physical interfaces there to.

It is *often* the analog components and housing that make things sound
suboptimal, and if you need proof of this, I call to your attention some
NPR phoners which are done with gear like the JK Audio BlueDriver 3, and
broadcast microphones. It's possible to get to within about 47% or so of
the sampling rate of the codec using gear like that, and it's pretty easy
(for a sound guy) to spot that combination in a live broadcast.

It's also worth noting that even if the recording format is VHS, it's very
easy to discern the difference between consumer cameras, pro SDTV, and pro
HDTV, in looking at the playback signal -- the differences are subtle, but
they are identifiable.

Now, those codecs *are* specially tuned for spoken word -- if you try
to stuff music down them, it's not gonna work very well at all...

Cheers,
-- jra
--
Jay R. Ashworth Baylink jra@baylink.com
Designer The Things I Think RFC 2100
Ashworth & Associates http://baylink.pitas.com 2000 Land Rover DII
St Petersburg FL USA http://photo.imageinc.us +1 727 647 1274
Re: Cellphones and Audio (was Ghost Click, though I got no idea why) [ In reply to ]
Jay Ashworth wrote:

> Now, those codecs *are* specially tuned for spoken word -- if you try
> to stuff music down them, it's not gonna work very well at all...

It was claimed to me many years ago that the 4kHz cutoff used in POTS
serves women and children less well than it does adult males. I have
never been aware that I have any greater problems understanding women or
children on the phone than I do men, but my hearing is not great. I
can't hear the difference between G.711 and G.729, for example, but some
people can.

Googling "PCM adult male voice", "4kHz adult male" and similar isn't
finding me anything. Was I told nonsense?
Re: Cellphones and Audio (was Ghost Click, though I got no idea why) [ In reply to ]
Adam Atkinson <ghira@mistral.co.uk> wrote;
> Jay Ashworth wrote:
> > Now, those codecs *are* specially tuned for spoken word -- if you try
> > to stuff music down them, it's not gonna work very well at all...
>
> It was claimed to me many years ago that the 4kHz cutoff used in POTS
> serves women and children less well than it does adult males. I have
> never been aware that I have any greater problems understanding women or
> children on the phone than I do men, but my hearing is not great. I
> can't hear the difference between G.711 and G.729, for example, but some
> people can.
>
> Googling "PCM adult male voice", "4kHz adult male" and similar isn't
> finding me anything. Was I told nonsense?


Probably. "sort of." <grin>

'Way back when', at least in the U.S., the 'voice' passband was 300-3000Hz.
Later, 300-3300Hz.

For perspective, rf you know anything about music, the 'A' below "Middle C'
is nominally 440Hz. 300Hz is roughly an octave below Middle C, and 3kHz is
2-1/2 octaves above it. That's the -high- end of the range for a piccolo,
or coloratura Soprano. Now, absent the overtones that give a note it's
'color', one of those high-pitch sources will sound more than a little bit
'tinny' over a classical 'voice passband' channel.

*HOWEVER*, the 'fundamental' frequencies for womens/childrens voices -is-
higher than that of adult males. But you're talking less than an octave
in 'most' cases. Less than 2 in 'extreme' (a guy with a _deep- bass voice
-- "basso profundo", and a 'squeaky' female/child) cases. This mean that
one does lose one to two additional 'overtones' of the fundamental on
women/children, vs. men.

This does, in general, *NOT* materially affect the 'intelligibility' of
the voice, although it does have a measurable adverse effect on the
'identifiability' of one such higher-pitched voice vis-a-vis a different
similarly-pitched voice. You lose more of the 'color' of their voices
vs the lower-pitched male voice.
Re: Cellphones and Audio (was Ghost Click, though I got no idea why) [ In reply to ]
----- Original Message -----
> From: "Adam Atkinson" <ghira@mistral.co.uk>

> Jay Ashworth wrote:
> > Now, those codecs *are* specially tuned for spoken word -- if you
> > try to stuff music down them, it's not gonna work very well at all...
>
> It was claimed to me many years ago that the 4kHz cutoff used in POTS
> serves women and children less well than it does adult males. I have
> never been aware that I have any greater problems understanding women
> or children on the phone than I do men, but my hearing is not great. I
> can't hear the difference between G.711 and G.729, for example, but
> some people can.
>
> Googling "PCM adult male voice", "4kHz adult male" and similar isn't
> finding me anything. Was I told nonsense?

No, you weren't. A 4khz channel is generally good from 3-400hz up to about
3.4khz, and if you look at spectrograms of the various categories of voices
you can see the differences, though they're not always as clear cut as you
might expect:

http://www.dplay.com/tutorial/bands/index.html

In general, though, intelligibility comes from the higher frequencies,
and 3.4kHz is *usually* high enough. What might be the case is that you'd
have more trouble *distinguishing* amongst women, or between women and
children, because the tones necessary for that are more located above the
cutoff frequency.

In short: it depends a lot on what you mean by 'serves well'. :-)

Cheers,
-- jra
--
Jay R. Ashworth Baylink jra@baylink.com
Designer The Things I Think RFC 2100
Ashworth & Associates http://baylink.pitas.com 2000 Land Rover DII
St Petersburg FL USA http://photo.imageinc.us +1 727 647 1274
Re: Cellphones and Audio (was Ghost Click, though I got no idea why) [ In reply to ]
On Thu, 03 May 2012 11:01:01 -0400, Jay Ashworth said:

> In general, though, intelligibility comes from the higher frequencies,
> and 3.4kHz is *usually* high enough. What might be the case is that you'd
> have more trouble *distinguishing* amongst women, or between women and
> children, because the tones necessary for that are more located above the
> cutoff frequency.

I have had more than a few surreal conversations on the phone with my
daughter - once the 3.4kHz filter gets done, I can't distinguish her voice from
her mom's (and yes, I've gotten social-engineered as a result). Life has
gotten simpler since she got old enough to have her own cell phone. ;)
Re: Cellphones and Audio (was Ghost Click, though I got no idea why) [ In reply to ]
Jay Ashworth wrote:

>> Googling "PCM adult male voice", "4kHz adult male" and similar isn't
>> finding me anything. Was I told nonsense?
>

[snippage]

> What might be the case is that you'd
> have more trouble *distinguishing* amongst women, or between women and
> children, because the tones necessary for that are more located above the
> cutoff frequency.

Thank you for this and the link. Very interesting stuff. I have never
tried to check to what extent I / others can distinguish different
female / young speakers on the phone. I shall try to pay more attention
to this in the future.

> In short: it depends a lot on what you mean by 'serves well'. :-)

Well, just the above seems like enough that you'd think there'd be more
(justified) grumbling that thanks to a choice made many many decades ago
it's harder to distinguish young or female speakers than it is adult
male ones. Maybe there is and I've just not noticed it. Is this one of
the things pushing adoption of higher bandwidth audio codecs? (My guess:
no.)
Re: Cellphones and Audio (was Ghost Click, though I got no idea why) [ In reply to ]
----- Original Message -----
> From: "Adam Atkinson" <ghira@mistral.co.uk>

> Well, just the above seems like enough that you'd think there'd be more
> (justified) grumbling that thanks to a choice made many many decades ago
> it's harder to distinguish young or female speakers than it is adult
> male ones. Maybe there is and I've just not noticed it. Is this one of
> the things pushing adoption of higher bandwidth audio codecs? (My guess:
> no.)

Not directly, I don't think, no. I suspect it's merely "why not?"

Cheers,
-- jra
--
Jay R. Ashworth Baylink jra@baylink.com
Designer The Things I Think RFC 2100
Ashworth & Associates http://baylink.pitas.com 2000 Land Rover DII
St Petersburg FL USA http://photo.imageinc.us +1 727 647 1274
Re: Cellphones and Audio (was Ghost Click, though I got no idea why) [ In reply to ]
On 5/3/12 10:29 , Jay Ashworth wrote:
> ----- Original Message -----
>> From: "Adam Atkinson" <ghira@mistral.co.uk>
>
>> Well, just the above seems like enough that you'd think there'd be more
>> (justified) grumbling that thanks to a choice made many many decades ago
>> it's harder to distinguish young or female speakers than it is adult
>> male ones. Maybe there is and I've just not noticed it. Is this one of
>> the things pushing adoption of higher bandwidth audio codecs? (My guess:
>> no.)
>
> Not directly, I don't think, no. I suspect it's merely "why not?"

wideband codecs carry music a lot better.

the can have considerably more dynamic range than you can expect from an
8 bit pcm mulaw encoding (about 45bB). that helps a lot in the speaker
phone situation.

if you have the opportunity to compare pstn and mp3 recordings of the
same meeting like I do on occasion the difference is considerable.


> Cheers,
> -- jra
RE: Cellphones and Audio (was Ghost Click, though I got no idea why) [ In reply to ]
As one involved in emergency services I don't gave a rats whether you
can't tell one voice from another. I do care if someone who is having a
fire, accident, cardiac episode or stroke can get through.

The cell companies are worrying about your whim and not the safety.



Ralph Brandt


-----Original Message-----
From: Valdis.Kletnieks@vt.edu [mailto:Valdis.Kletnieks@vt.edu]
Sent: Thursday, May 03, 2012 11:33 AM
To: NANOG
Subject: Re: Cellphones and Audio (was Ghost Click, though I got no idea
why)

On Thu, 03 May 2012 11:01:01 -0400, Jay Ashworth said:

> In general, though, intelligibility comes from the higher frequencies,
> and 3.4kHz is *usually* high enough. What might be the case is that
you'd
> have more trouble *distinguishing* amongst women, or between women and
> children, because the tones necessary for that are more located above
the
> cutoff frequency.

I have had more than a few surreal conversations on the phone with my
daughter - once the 3.4kHz filter gets done, I can't distinguish her
voice from
her mom's (and yes, I've gotten social-engineered as a result). Life
has
gotten simpler since she got old enough to have her own cell phone. ;)
RE: Cellphones and Audio (was Ghost Click, though I got no idea why) [ In reply to ]
I wanted to mention one other thing here. In addition to my day job I
am a ham radio operator and a IMT COML. I deal with UNDERSTANDABILITY.
I sympathize with you on getting bamboozled by the family, have 3
sisters and two daughters.

But in the real world of communications 2.3 - 2.5 Khz filters for SSB
are the norm, and some guys like it tighter.

Let me be frank, spectrum space and hence bandwidth is finite. There is
no silver bullet. The FCC mandated narrow banding in VHF is costing
public service millions. AT&T pronouncements aside, I have heard them,
there will be improvements but not quantum ones. Right now the cell cos
are trying to desperately get more spectrum because they know they are
about to hit the wall. They already do in places like unHappy Valley
(Penn State) at game time when everyone is on the cells.

One recent merger/acquisition attempt that failed was for the larger co
to get the smaller one's spectrum space. They are looking at some
public service spectrum to see if they can offer enough money to get it
- most of that money is spent lobbying. Once it goes to cell, it isn't
coming back no matter what. The Digital TV was as much about freeing
spectrum space to sell it for government income as better TV signals
which we didn't need and have cost every person in the US at least
$1,000 (new TV's, increased ad costs which increase product prices to
pay for the broadcasting equipment) and we didn't get that back for the
spectrum space when it was sold. We paid for the industry to upgrade!




Ralph Brandt


-----Original Message-----
From: Valdis.Kletnieks@vt.edu [mailto:Valdis.Kletnieks@vt.edu]
Sent: Thursday, May 03, 2012 11:33 AM
To: NANOG
Subject: Re: Cellphones and Audio (was Ghost Click, though I got no idea
why)

On Thu, 03 May 2012 11:01:01 -0400, Jay Ashworth said:

> In general, though, intelligibility comes from the higher frequencies,
> and 3.4kHz is *usually* high enough. What might be the case is that
you'd
> have more trouble *distinguishing* amongst women, or between women and
> children, because the tones necessary for that are more located above
the
> cutoff frequency.

I have had more than a few surreal conversations on the phone with my
daughter - once the 3.4kHz filter gets done, I can't distinguish her
voice from
her mom's (and yes, I've gotten social-engineered as a result). Life
has
gotten simpler since she got old enough to have her own cell phone. ;)