K3 Audio Monitor Delay

[Tom Boucher]

Lyle,

Thanks for the explanation. As with most hams, I like the sound of my own voice (!) and listening to a delayed version of my own voice on phones during a long contest, I find quite wearing. Would it be possible to menu-select the monitor output prior to any of the processing you mention? I've reverted to using my dear old TS-850 for phone operation because of this problem.

73
Tom G3OLB

Lyle wrote:

The delay is due to: audio input anti-aliasing filter in the A to D, TX
AF filtering, phase shift network, interpolation filter (conversion from
12 kHz sampling rate to 48 kHz sampling rate) , post-IF clipper filter,
decimation filter (conversion from 48 kHz sampling rate to 12 kHz
sampling rate), reconstruction filter (part of the D to A).

We thought it best to allow you to hear the effects of the IF signal
processing, so delays are unavoidable.  Most people do not notice as the
K3 uses filters optimized to minimize delays.  Note that Tx EQ adds zero
additional delay. Like the Rx EQ, the Tx EQ just adds more constraints
to the AF FIR filter generator.

There may be opportunity to further redice delays by combining some
filter functions.  One of the items on my list of background tasks :-)

______________________________________________________________
Elecraft mailing list
Home: http://mailman.qth.net/mailman/listinfo/elecraft
Help: http://mailman.qth.net/mmfaq.htm
Post: mailto:[hidden email]

This list hosted by: http://www.qsl.net
Please help support this email list: http://www.qsl.net/donate.html

[Jack Smith-6]

Let me add some quantified data on monitoring delay.

In  the telephone industry, the signal from the transmitter back to the
earpiece is is called "sidetone." In the conventional analog telephone
network, sidetone is generated locally in the telephone instrument, but
can also come from reflections in the network, called echo, often
occurring in 2-wire to 4-wire transitions. Echo suppressors are employed
to reduce the reflections because it was quickly learned that delayed
echos are disconcerting to telephone users.

With the replacement of analog transmission to digital technology, and
in particular speech compression and transcoding required in digital
mobile telephone systems, side tone delay and echo control has again
become something that equipment and network designers must consider.

In any event, the effect of delayed sidetone upon articulation by the
speaker has long been studied, and the generally accepted view is that
anything more than 50 ms is cause for concern and values exceeding 100
ms are a red flag. (A related topic is overall delay, which causes
uncertainty over when one speaker stops and other begins. This is a real
problem in tandem satellite links, and even more so when a digital
mobile radio system with transcoding and transmission delays are
present, to the point where digital mobile telephone switching
algorithms usually set the "no satellite flag" to yes.)

An excellent summary of the typical research can be found at
http://scitation.aip.org/getabs/servlet/GetabsServlet?prog=normal&id=JASMAN0000650000S100S115000006&idtype=cvips&gifs=yes&ref=no,
reproduced  below.

    "Sidetone^ delay (delayed auditory feedback) is known to cause
    speakers to^ talk more slowly, and at delays of about 200 ms^ also
    causes severe articulation disturbances. If shorter delays can slow^
    speech without adverse effects on articulation, this effect could
    be^ used to make talkers speak more slowly in certain
    communications^ situations. Subjects read lists of 1, 2, 3, or 4^
    syllable words and of sentences under each of 6 delay^ conditions
    (0, 7.5, 15, 30, 60, and 100 ms). Delays^ of 15--30 ms caused a
    significant slowdown in speaking rate^ with no adverse effects on
    articulation. Delays of 60--100 ms^ caused a greater slowdown but
    also had a noticeable adverse^ effect on speech quality, especially
    for sentences and polysyllablic lists.^ In communications situations
    where the quality of transmitted speech is^ degraded (e.g., vocoded
    speech), sidetone delay may be useful in^ modifying the talk's
    behavior to improve overall speech quality."

Most studies use general population research subjects, but skilled and
experienced operators (amateur radio or military) may be able to
tolerate greater delays without harmful effect.

The general user data suggests that radio monitor delay should be 50 ms
or less, with the emphasis on "less."

Anyone interested in researching the subject should Google "sidetone
delay" or "delayed auditory feedback" and you will find more about the
subject than you ever wanted to know.

Jack K8ZOA
www. cliftonlaboratories.com
______________________________________________________________
Elecraft mailing list
Home: http://mailman.qth.net/mailman/listinfo/elecraft
Help: http://mailman.qth.net/mmfaq.htm
Post: mailto:[hidden email]

This list hosted by: http://www.qsl.net
Please help support this email list: http://www.qsl.net/donate.html

[Stewart Baker]

Very interesting Jack.
I might get round to injecting an audio tone into the mic input
and measure the delay to it appearing at the TX monitor.

73
Stewart G3RXQ
On Sat, 16 Jan 2010 07:12:07 -0500, Jack Smith wrote:
> Let me add some quantified data on monitoring delay.
>
> In  the telephone industry, the signal from the transmitter back
to the
> earpiece is is called "sidetone." In the conventional analog
telephone
> network, sidetone is generated locally in the telephone
instrument, but
> can also come from reflections in the network, called echo,
often
> occurring in 2-wire to 4-wire transitions. Echo suppressors are
employed
> to reduce the reflections because it was quickly learned that
delayed
> echos are disconcerting to telephone users.
>
> With the replacement of analog transmission to digital
technology, and
> in particular speech compression and transcoding required in
digital
> mobile telephone systems, side tone delay and echo control has
again
> become something that equipment and network designers must
consider.
>
> In any event, the effect of delayed sidetone upon articulation
by the
> speaker has long been studied, and the generally accepted view
is that
> anything more than 50 ms is cause for concern and values
exceeding 100
> ms are a red flag. (A related topic is overall delay, which
causes
> uncertainty over when one speaker stops and other begins. This
is a real
> problem in tandem satellite links, and even more so when a
digital
> mobile radio system with transcoding and transmission delays are
> present, to the point where digital mobile telephone switching
> algorithms usually set the "no satellite flag" to yes.)
>
> An excellent summary of the typical research can be found at
> http://scitation.aip.org/getabs/servlet/GetabsServlet?prog=norma
l&id=JASMAN0000650000S100S115000006&idtype=cvips&gifs=yes&ref=no,
> reproduced  below.
>
>    "Sidetone^ delay (delayed auditory feedback) is known to
cause
>    speakers to^ talk more slowly, and at delays of about 200 ms^
also
>    causes severe articulation disturbances. If shorter delays
can slow^
>    speech without adverse effects on articulation, this effect
could
>    be^ used to make talkers speak more slowly in certain
>    communications^ situations. Subjects read lists of 1, 2, 3,
or 4^
>    syllable words and of sentences under each of 6 delay^
conditions
>    (0, 7.5, 15, 30, 60, and 100 ms). Delays^ of 15--30 ms caused
a
>    significant slowdown in speaking rate^ with no adverse
effects on
>    articulation. Delays of 60--100 ms^ caused a greater slowdown
but
>    also had a noticeable adverse^ effect on speech quality,
especially
>    for sentences and polysyllablic lists.^ In communications
situations
>    where the quality of transmitted speech is^ degraded (e.g.,
vocoded
>    speech), sidetone delay may be useful in^ modifying the
talk's
>    behavior to improve overall speech quality."
>
> Most studies use general population research subjects, but
skilled and
> experienced operators (amateur radio or military) may be able to
> tolerate greater delays without harmful effect.
>
> The general user data suggests that radio monitor delay should
be 50 ms
> or less, with the emphasis on "less."
>
> Anyone interested in researching the subject should Google
"sidetone
> delay" or "delayed auditory feedback" and you will find more
about the

> subject than you ever wanted to know.
>
> Jack K8ZOA
> www. cliftonlaboratories.com
> ______________________________________________________________
> Elecraft mailing list
> Home: http://mailman.qth.net/mailman/listinfo/elecraft
> Help: http://mailman.qth.net/mmfaq.htm
> Post: mailto:[hidden email]
>
> This list hosted by: http://www.qsl.net
> Please help support this email list:

http://www.qsl.net/donate.html


______________________________________________________________
Elecraft mailing list
Home: http://mailman.qth.net/mailman/listinfo/elecraft
Help: http://mailman.qth.net/mmfaq.htm
Post: mailto:[hidden email]

This list hosted by: http://www.qsl.net
Please help support this email list: http://www.qsl.net/donate.html

[Mike Harris]

Hi,

Similarly, try injecting an RF signal into the antenna jack and see
how long it takes to show at the AF, or vanish when the signal is
removed.  You will find it interesting.  Ultra contesters should
stick to analogue if they want nano second responsiveness to keep
their rate up.

Personally I don't take any notice of the monitor audio other than
it offers a certain presence.  The original function of sidetone in
telephones was to stop people shouting and it works very well for
that in the K3.  That this is effective is demonstrated by the loud
half conversations heard when in the presence of mobile phone users.
The presence of small mirrors in UK phone boxes many years ago was a
behaviour modifier.  Users could see a face in the box and behaved
as if they were being observed.

Analogue echo suppressors have long been replaced with digital echo
cancellers, much more effective.

Regards,

Mike VP8NO

______________________________________________________________
Elecraft mailing list
Home: http://mailman.qth.net/mailman/listinfo/elecraft
Help: http://mailman.qth.net/mmfaq.htm
Post: mailto:[hidden email]

This list hosted by: http://www.qsl.net
Please help support this email list: http://www.qsl.net/donate.html

[P.B. Christensen]

> anything more than 50 ms is cause for concern and values exceeding 100
> ms are a red flag. (A related topic is overall delay, which causes
> uncertainty over when one speaker stops and other begins. This is a real
> problem in tandem satellite links..

Typical 80 ms of latency over the Internet is enough to completely render
ear-to-hand coordination with an iambic paddle as "impossible."  The latency
is effectively doubled when one considers that it may take 80 ms for the key
closure to arrive at the transmitter but it takes another 80 ms for the
transmitter's sidetone to reach the operator at the remote location for a
net delay of ~ 160 ms.  To the best of my knowledge, this type of "dual
latency" with CW has not been studied in depth.  Of course, latencies are
dependent on the IP path.  When operating from my hotel in Italy last year,
latency was approximately double the worst latencies I've measured in NA.

When using a manual key or paddle, it necessitates the use of a local
sidetone injected into the headphone audio.  Some of the QSK attributes are
still possible, only Rx audio between keyed elements appear delayed in the
headphones.  That's not a problem so long as the locally-injected side-tone
is synchronized with the sending.

By contrast, I've found when using a CW keyboard from my netbook at a remote
location, latency is of no consequence since the brain is typing as if
writing a paper but the brain does not require audio feedback from the
sidetone to keep the typing synchronized.

Paul, W9AC

______________________________________________________________
Elecraft mailing list
Home: http://mailman.qth.net/mailman/listinfo/elecraft
Help: http://mailman.qth.net/mmfaq.htm
Post: mailto:[hidden email]

This list hosted by: http://www.qsl.net
Please help support this email list: http://www.qsl.net/donate.html

[Jim Brown-10]

On Sat, 16 Jan 2010 10:02:49 -0500, Paul Christensen wrote:


>When using a manual key or paddle, it necessitates the use of a local
>sidetone injected into the headphone audio.

The same thing happens when a live talker (or singer) tries to use a mic
on the field in a stadium (or sing the anthem) -- you hear own voice
delayed by the acoustic travel time between the loudspeaker system and
you, plus any DSP delay (typically 5-15msec), plus any delay used for
synchronization of loudspeakers (typically 1-150 msec). The total delay
can easily reach 120-150msec in a professional sports venue. VERY few
people can sing or talk against that delay. Long ago, we learned to solve
that problem just as you have outlined -- the singer/talker is fed an
undelayed signal from his/her mic, plus any musical accompaniment, via
either a local loudspeaker (floor monitor) or an earphone.  

73,

Jim Brown K9YC


______________________________________________________________
Elecraft mailing list
Home: http://mailman.qth.net/mailman/listinfo/elecraft
Help: http://mailman.qth.net/mmfaq.htm
Post: mailto:[hidden email]

This list hosted by: http://www.qsl.net
Please help support this email list: http://www.qsl.net/donate.html

[Joe Subich, W4TV-4]

The delayed sidetone issues are well known in broadcast news
circles.  They first appeared with satellite circuits where
the > 250 msec delay gave even the most accomplished speaker
a stutter.  The problem became even more acute when field
cueing migrated from analog to digital cell phones and the
digital processing delis created >100 msec delays on even
"local" circuits.  The problem has become even more acute in
the last few years as the microwave equipment used for (local)
field acquisition has transitioned from analog to digital and
added yet one more layer of delays.  

In each case, the solution has been to provide mix-minus
(program audio without the specific remote site audio) and
add locally generated sidetone at the remote site.  While
that works, managing a unique mix-minus feed for each source
(location) in a multi-remote program - particularly when
"talent" and/or guests at the remote sites must interact -
is not exactly simple <G>.

73,

   ... Joe, W4TV
 

> -----Original Message-----
> From: [hidden email]
> [mailto:[hidden email]] On Behalf Of Jack Smith
> Sent: Saturday, January 16, 2010 7:12 AM
> To: [hidden email]
> Subject: [Elecraft] K3 Audio Monitor Delay
>
>
> Let me add some quantified data on monitoring delay.
>
> In  the telephone industry, the signal from the transmitter
> back to the
> earpiece is is called "sidetone." In the conventional analog
> telephone
> network, sidetone is generated locally in the telephone
> instrument, but
> can also come from reflections in the network, called echo, often
> occurring in 2-wire to 4-wire transitions. Echo suppressors
> are employed
> to reduce the reflections because it was quickly learned that delayed
> echos are disconcerting to telephone users.
>
> With the replacement of analog transmission to digital
> technology, and
> in particular speech compression and transcoding required in digital
> mobile telephone systems, side tone delay and echo control has again
> become something that equipment and network designers must consider.
>
> In any event, the effect of delayed sidetone upon articulation by the
> speaker has long been studied, and the generally accepted
> view is that
> anything more than 50 ms is cause for concern and values
> exceeding 100
> ms are a red flag. (A related topic is overall delay, which causes
> uncertainty over when one speaker stops and other begins.
> This is a real
> problem in tandem satellite links, and even more so when a digital
> mobile radio system with transcoding and transmission delays are
> present, to the point where digital mobile telephone switching
> algorithms usually set the "no satellite flag" to yes.)
>
> An excellent summary of the typical research can be found at
> http://scitation.aip.org/getabs/servlet/GetabsServlet?prog=nor

mal&id=JASMAN0000650000S100S115000006> &idtype=cvips&gifs=yes&ref=no,

> reproduced  below.
>
>     "Sidetone^ delay (delayed auditory feedback) is known to cause
>     speakers to^ talk more slowly, and at delays of about 200 ms^ also
>     causes severe articulation disturbances. If shorter
> delays can slow^
>     speech without adverse effects on articulation, this effect could
>     be^ used to make talkers speak more slowly in certain
>     communications^ situations. Subjects read lists of 1, 2, 3, or 4^
>     syllable words and of sentences under each of 6 delay^ conditions
>     (0, 7.5, 15, 30, 60, and 100 ms). Delays^ of 15--30 ms caused a
>     significant slowdown in speaking rate^ with no adverse effects on
>     articulation. Delays of 60--100 ms^ caused a greater slowdown but
>     also had a noticeable adverse^ effect on speech quality,
> especially
>     for sentences and polysyllablic lists.^ In communications
> situations
>     where the quality of transmitted speech is^ degraded
> (e.g., vocoded
>     speech), sidetone delay may be useful in^ modifying the talk's
>     behavior to improve overall speech quality."
>
> Most studies use general population research subjects, but
> skilled and
> experienced operators (amateur radio or military) may be able to
> tolerate greater delays without harmful effect.
>
> The general user data suggests that radio monitor delay
> should be 50 ms
> or less, with the emphasis on "less."
>
> Anyone interested in researching the subject should Google "sidetone
> delay" or "delayed auditory feedback" and you will find more
> about the
> subject than you ever wanted to know.
>
> Jack K8ZOA
> www. cliftonlaboratories.com
> ______________________________________________________________
> Elecraft mailing list
> Home: http://mailman.qth.net/mailman/listinfo/elecraft
> Help: http://mailman.qth.net/mmfaq.htm
> Post: mailto:[hidden email]
>
> This list hosted by: http://www.qsl.net
> Please help support this email list: http://www.qsl.net/donate.html

______________________________________________________________
Elecraft mailing list
Home: http://mailman.qth.net/mailman/listinfo/elecraft
Help: http://mailman.qth.net/mmfaq.htm
Post: mailto:[hidden email]

This list hosted by: http://www.qsl.net
Please help support this email list: http://www.qsl.net/donate.html