Latency

Windows 7 is infamous for latency issues which you can hear demonstrated
on the 'net as windows 7 latency, and which I have personally
experienced. Windows XP did not have this problem.

Does anyone know if Microsoft has eliminated latency in 10?

Thanks.

riserman

On Sat, 02 May 2015 12:29:50 +0000, Stormin' Norman
wrote:

On Fri, 01 May 2015 20:30:30 -0400, riserman wrote:

latency issues which you can hear

What kind of latency are you speaking of?

http://lmgtfy.com/?q=hear+windows+7+latency

On 05/01/2015 07:30 PM, riserman wrote:
Windows 7 is infamous for latency issues which you can hear demonstrated
on the 'net as windows 7 latency, and which I have personally
experienced. Windows XP did not have this problem.

Does anyone know if Microsoft has eliminated latency in 10?

Thanks.

riserman



I use both XP and Win7 on occasion and have not seen any latency on Win7


If you want help. give specifics

Michael Logies wrote:
On Sat, 02 May 2015 12:29:50 +0000, Stormin' Norman
wrote:

On Fri, 01 May 2015 20:30:30 -0400, riserman wrote:

latency issues which you can hear

What kind of latency are you speaking of?

http://lmgtfy.com/?q=hear+windows+7+latency

Stormin' is playing the "20 questions game".
Let the game proceed as it may...

http://en.wikipedia.org/wiki/Twenty_questions

You could throw in a "is it bigger than a bread box"
at the appropriate moment. But first we have to see
whether the OP wants to play or not.

Paul

Stormin' Norman wrote:
On Sat, 02 May 2015 10:24:40 -0400, Paul wrote:

Michael Logies wrote:
On Sat, 02 May 2015 12:29:50 +0000, Stormin' Norman
wrote:

On Fri, 01 May 2015 20:30:30 -0400, riserman wrote:
latency issues which you can hear
What kind of latency are you speaking of?
http://lmgtfy.com/?q=hear+windows+7+latency
Stormin' is playing the "20 questions game".
Let the game proceed as it may...


You are right, it was silly of me to ask for clarification.

Someone has to do it, and it was your turn.

For low latency, there is a thread here with an example.
And the people use ASIO.

https://social.technet.microsoft.com...m=w7itpromedia

It might be the delay between video in a movie and
the audio.

Or it might be latency as measured by DPC_LAT,
which causes problem for audio workstation owners
(a motherboard problem).

Or it could be the impact of adding echo suppression
in the form of an in-game audio feature (talk using
headset while gaming).

OK, so "is it bigger than a bread box" is my next question.

Paul

On 5/1/2015 8:30 PM, riserman wrote:
Windows 7 is infamous for latency issues which you can hear demonstrated
on the 'net as windows 7 latency, and which I have personally
experienced. Windows XP did not have this problem.

Does anyone know if Microsoft has eliminated latency in 10?

Thanks.

riserman
To All of You,

I'm talking about audio latency. The discussion group reports all sorts
of attempts to overcome this echo effect, but not many solutions I can
use. For an excellent example of the problem, try

www.youtube.com/watch?v=d7sc-kl7bf0

Thanks for your suggestions. I've tried them all without good result.

riserman

On 5/2/2015 12:50 PM, riserman wrote:
On 5/1/2015 8:30 PM, riserman wrote:
Windows 7 is infamous for latency issues which you can hear demonstrated
on the 'net as windows 7 latency, and which I have personally
experienced. Windows XP did not have this problem.

Does anyone know if Microsoft has eliminated latency in 10?

Thanks.

riserman
To All of You,

I'm talking about audio latency. The discussion group reports all sorts
of attempts to overcome this echo effect, but not many solutions I can
use. For an excellent example of the problem, try

www.youtube.com/watch?v=d7sc-kl7bf0

Thanks for your suggestions. I've tried them all without good result.

riserman

What I forgot to say was does this audio latency problem occur with
Windows 10. Has Microsoft solved the problem?

Thanks,

riserman

In message , riserman
writes
I'm talking about audio latency. The discussion group reports all sorts
of attempts to overcome this echo effect, but not many solutions I can
use. For an excellent example of the problem, try

www.youtube.com/watch?v=d7sc-kl7bf0

Thanks for your suggestions. I've tried them all without good result.

riserman

What I forgot to say was does this audio latency problem occur with
Windows 10. Has Microsoft solved the problem?

I suspect that the latency in Windows 7, 8 and perhaps 10 is a
combination of :
1. The audio interfaces have features that were not present in most XP
generation machines. HD Audio works over an internal serial connection
to the mainboard audio system. Many systems now have "effects" that are
useless, but probably introduce delay while processing.

2. The later Microsoft audio protocol and mixing application depends on
its ability to convert the sample rate to the rate selected in the
Windows mixer.
In Windows7 this caused distortion
http://support.microsoft.com/kb/2653312
and I would expect this converter to contribute latency.
The conversion of everything to one sample rate allows the newer Windows
mixer to mix audio of different types and from different sources.
I don't know whether the Microsoft patch is needed in 8 or 10 as well.

3. Proper audio interfaces use asio to bypass the Windows system and
good applications** can apply an adjustment to set the latency to zero.

4. Proper audio interfaces have a direct audio output on the device
itself rather than pratting about with the Windows mixer that can never
be all things to all men (or women).

The video referred to by the OP is an example of the effect, but does
nothing to describe how the machine and Windows is set up for audio. For
example, my experience is that out of the box Windows comes set up for
different sample rates for record and play, which always seems to me
just stupid.

** Adobe Audition 3 with an Edirol UA-4FX that I use actually measures
as having a negative latency in asio of a couple of cycles.

--
Bill

riserman wrote:
On 5/1/2015 8:30 PM, riserman wrote:
Windows 7 is infamous for latency issues which you can hear demonstrated
on the 'net as windows 7 latency, and which I have personally
experienced. Windows XP did not have this problem.

Does anyone know if Microsoft has eliminated latency in 10?

Thanks.

riserman
To All of You,

I'm talking about audio latency. The discussion group reports all sorts
of attempts to overcome this echo effect, but not many solutions I can
use. For an excellent example of the problem, try

www.youtube.com/watch?v=d7sc-kl7bf0

Thanks for your suggestions. I've tried them all without good result.

riserman

So the word "latency" isn't even part of the problem.
That's an "echo" problem.

The person who made that ten minute rant on Youtube,
already demonstrated a path for it. If you enable the
"monitor" function, that's the same as the "What You Hear"
audio path of old. It loops the output back to the input,
and is a preferred method of old, to record movie output.
Say a movie output is AC3 encoded, you wait for the movie
software to decode it for you, then record it off the
stereo output. (I don't think there are enough
input channels on motherboard audio, to get all the channels
at once, like 5.1 maybe.)

I used to have a problem like that on WinXP, and it might
have been a CMedia card driver that was doing it. The control
panel for the driver had a "digital effects" menu. One
of the effects would be "concert hall", which adds reverberation
with a 30 millisecond time constant. I have a relatively
"tin ear", and I noticed what little music I played
on the computer, sounded like crap. It sounded "mushy" and
garbled.

So I did an experiment with two sound programs. On one sound
program, I "drew" a waveform, an impulse signal which you
would hear as a tick or pop. With a second program I recorded
the output. And compared the recorded signal to the original
signal. Now, I did not expect the returned signal to be a
perfect match. The bandpass filter on input, prevents steep
walled signals (bandwidth above Nyquist frequency) from
getting through. But when I looked at the result, I could
see this. And this is with "Concert Hall" effect type
set to "None".

/\
/ \ /\
___/ \___________/ \______________/\____

--- 30msec --- --- 30msec ---

So that was a driver issue, and was occurring on the
"perfect" WinXP. A comparison with a different brand
of sound chip, with the effects on that device set to
"None", actually had no effect at all.

On the one hand, you can "add reverberation" to a signal.

On the other hand, you can use an "echo suppressor" to remove it.
Echo suppressors are used for in-game talk programs,
and some of those suppressors run full time (even when
no game is being played). The echo suppressors can even
prevent a 5.1 card from entering 5.1 mode, and force
the card to stay in stereo mode.

http://en.wikipedia.org/wiki/Echo_suppressor

"The concept of an echo canceller is to synthesize an
estimate of the echo from the talker's signal, and subtract
that synthesis from the return path instead of switching
attenuation into/out of the path. This technique requires
adaptive signal processing to generate a signal accurate
enough to effectively cancel the echo, where the echo can
differ from the original due to various kinds of degradation
along the way."

So that's how you'd remove something, that shouldn't have
been there in the first place.

*******

Latency, is a measure of processing delay. Some audio workstation
users use ASIO drivers and non-gaming sound cards, to achieve
low millisecond delay from when the computer emits a digital
sample, until it becomes analog on the ASIO hardware ourput.

Latency is also used in a deterministic way. For example,
some movie players have necessary latency to decode a compressed
audio signal. the latency is padded out to half a second,
and as long as the "delay variance" is low, the movie
player simply delays the presentation of the video frames
by half a second too, giving good lip sync. The trick there
is not the latency, rather it's a matter of having low
delay variance and padding things with the appropriate static
delay.

The lowest delay AC3 decoding, might have been on my Nforce2
chipset motherboard, where the Southbridge had five little DSP
processors. Some of which had dedicated functions, while
others could be programmed for different uses. The AC3 was
done in there, rather than the main processor. But even at
50 milliseconds, that still might require padding and
compensation, to get proper lip sync.

But the properly shown in the Youtube movie is "echo" or
"artificial reverberation", and it can be added by an
audio driver, and it can happen even when the "Effects"
control is set to "None". There's nothing you can do
about that, short of using an echo suppressor/canceller.

*******

So you need to look for a model of the audio stack for
each OS, and decide if there is something about the stack
that can be amiss. And all the ingredients are here.

http://forums.guru3d.com/showthread.php?t=351061

The claim in the article copied here, is there is a loopback
path for controlling echo. But that loopback path is for
echo cancellation, rather than for "What you hear". The
What You Hear or Stereo Mix feature, would be outside this
block.

http://img339.imageshack.us/img339/3291/49255217.png

And even this diagram doesn't have Stereo Mix. No features
of the audio input stack are shown.

http://img836.imageshack.us/img836/2505/96320145.png

Now, say, instead of implementing what is in the diagram,
somehow the developers end up with *two* loopback paths.
And inappropriate local processing is done in both chunks
of the diagram.

It's a recipe for echo. It requires someone overseeing the
development, to catch implementation failures. And if the
sound card company is just looking at their own development,
maybe they're missing an interaction with the software engine.

http://stackoverflow.com/questions/5...indows-vista-7

This incident, would be different, as the time constant is longer.
Starting at page 32, I can see the problem is still not fixed.
That means it's a DRM-induced failure in the architecture (i.e.
WontFix because of DRM and preventing exact audio copy).

http://answers.microsoft.com/en-us/w...b-f0cdc78a702d

My test machine is down for maintenance for a while, so fiddling
with this will have to wait for later. The test machine has
Windows 7 on it.

Funny suggestion here. As if this is going to happen...

https://support.microsoft.com/en-us/kb/2577040/

"You may also be able to reduce the feedback loop by
selecting the Enhancement tab in the microphone's properties
page and enabling one of the following options:

Enable Noise Suppression
Enable Acoustic Echo Cancellation

Note: These settings are audio hardware and software
specific and may not be available for all microphones.
"

I did manage to find one picture. And it even has RealTek in the name.

http://l8pd2nvxwokyuct7159r.cdn01.so...timization.jpg

(from http://blog.layer8.sh/2012/07/12/060...oft-windows-7/ )

http://www.techspot.com/community/to...effect.178881/

"You may have a sample rate problem. I had the same echoey
issue and mine was resolved with making sure the windows
sample rate was the same as the Realtek HD sample rate.

Both should be set to 16bit 48000Hz. Go to

Control Panel Sound Speakers (Realtek Audio Device) Properties Advanced

then select the above sample rate from the drop-down menu
and apply. In the Realtek Menu (click on speaker icon in
bottom right of Windows) Select the Default Format tab then
make sure the same sample rate is listed in that drop-down menu.
"

There's a leakage path of some sort.

http://forum.teamspeak.com/showthrea...audio-recorder)

I'll try this out later on the test machine, to see
if I have the same symptoms. I have some headpnones
I can use for the test (to rule out a feedback path).

Paul

riserman wrote:
On 5/1/2015 8:30 PM, riserman wrote:
Windows 7 is infamous for latency issues which you can hear demonstrated
on the 'net as windows 7 latency, and which I have personally
experienced. Windows XP did not have this problem.

Does anyone know if Microsoft has eliminated latency in 10?

Thanks.

riserman
To All of You,

I'm talking about audio latency. The discussion group reports all sorts
of attempts to overcome this echo effect, but not many solutions I can
use. For an excellent example of the problem, try

www.youtube.com/watch?v=d7sc-kl7bf0

Thanks for your suggestions. I've tried them all without good result.

riserman

Well, I tested here. Put my good microphone on the line in and
microphone ports, and did two separate short recordings. No echo.
Win 7 SP1 Home Premium, RealTek audio in stereo mode.

The reason I used Line In, is the microphone has a line level
output. It has a tiny four pin amplifier chip inside the
microphone, and the mike runs off +5V. But I also tried it on
the microphone port, and there was no echo in either recording.

And while I threatened to use headphones, for a first test
I used speakers. And when in "listen" mode, I could hear
the delay from spoken word, to output from speaker. But
once switched out of "listen", and attempting a recording,
everything worked fine.

So maybe there is software that I'm not using, which is doing it.

I tested with Audacity (fresh download, Version 2.1), as well as
SoundRecorder in Windows.

That OS is relatively free of cruft, as it was a test
OS for helping someone else, and only the items to be
tested were added (SilverLight, Flash, Firefox, Avast AV).

Paul

In message , Paul writes
The person who made that ten minute rant on Youtube,
already demonstrated a path for it. If you enable the
"monitor" function, that's the same as the "What You Hear"
audio path of old. It loops the output back to the input,
and is a preferred method of old, to record movie output.
Say a movie output is AC3 encoded, you wait for the movie
software to decode it for you, then record it off the
stereo output. (I don't think there are enough
input channels on motherboard audio, to get all the channels
at once, like 5.1 maybe.)

Paul, I'm afraid I don't follow this (or some of the stuff I've
snipped!).

The man on YouTube has "Stereo Mix" (aka "What You Hear"), which is what
you are referring to, disabled.

It is different , where it exists, in the way it works in XP from the
way it works in 7.8 and 10 (I think, but haven't got a 10 test machine
with stereo Mix).
It takes the output of the playback sound faders and presents this as an
input on the recording mixer.
In XP, the mix to the "What you hear" is a separate mix taken from just
before the playback faders, so you can record from it with the playback
faders faded down. In 7 & 8, the mix is taken after the playback faders,
so fading the playback controls down affects what is recorded.

This is different from the "Listen" function on the Microphone channel,
which is taking the audio from after the recording fader in the
recording mixer (if there is one - depends on the sound card) and
feeding it to the playback destination. I.e. the endpoint, not the
playback mixer.

I think the other things you pointed to described all sorts of things
that were not strictly related. One describes an audio architecture
that, aiui, never actually happened, but looks like a Microsoft
suggestion of a parallel implementation of something like asio.

I can't find the demos I once made, but they showed Adobe Audition 3, a
multitrack editor, connected to an Edirol UA-4FX with its output linked
physically to its input, This combination can use asio drivers or
Microsoft DirectX drivers. I can edit down an audio sine wave to leave
just a handful of cycles, play it on one track and record it on another
track, then show the tracks together to measure the time difference.

DirectX shows a small delay, ASIO showed no delay.

The audio systems in laptops have all sorts of processing for echo
cancellation, noise suppression etc. Noise suppression can sound
terrible and is, I think, enabled by default on many cheap laptops
because some idiot designer has put the internal microphone right next
to the HD and fan.

All this processing must take time. I don't know whether the signal goes
through the processing hardware or software if processing is switched
out, but I wouldn't be surprised.

Realtek used to have schematics and block diagrams on their site. Not
sure if they still do.
--
Bill

Bill wrote:
In message , Paul writes
The person who made that ten minute rant on Youtube,
already demonstrated a path for it. If you enable the
"monitor" function, that's the same as the "What You Hear"
audio path of old. It loops the output back to the input,
and is a preferred method of old, to record movie output.
Say a movie output is AC3 encoded, you wait for the movie
software to decode it for you, then record it off the
stereo output. (I don't think there are enough
input channels on motherboard audio, to get all the channels
at once, like 5.1 maybe.)

Paul, I'm afraid I don't follow this (or some of the stuff I've snipped!).

The man on YouTube has "Stereo Mix" (aka "What You Hear"), which is what
you are referring to, disabled.

It is different , where it exists, in the way it works in XP from the
way it works in 7.8 and 10 (I think, but haven't got a 10 test machine
with stereo Mix).
It takes the output of the playback sound faders and presents this as an
input on the recording mixer.
In XP, the mix to the "What you hear" is a separate mix taken from just
before the playback faders, so you can record from it with the playback
faders faded down. In 7 & 8, the mix is taken after the playback faders,
so fading the playback controls down affects what is recorded.

This is different from the "Listen" function on the Microphone channel,
which is taking the audio from after the recording fader in the
recording mixer (if there is one - depends on the sound card) and
feeding it to the playback destination. I.e. the endpoint, not the
playback mixer.

I think the other things you pointed to described all sorts of things
that were not strictly related. One describes an audio architecture
that, aiui, never actually happened, but looks like a Microsoft
suggestion of a parallel implementation of something like asio.

I can't find the demos I once made, but they showed Adobe Audition 3, a
multitrack editor, connected to an Edirol UA-4FX with its output linked
physically to its input, This combination can use asio drivers or
Microsoft DirectX drivers. I can edit down an audio sine wave to leave
just a handful of cycles, play it on one track and record it on another
track, then show the tracks together to measure the time difference.

DirectX shows a small delay, ASIO showed no delay.

The audio systems in laptops have all sorts of processing for echo
cancellation, noise suppression etc. Noise suppression can sound
terrible and is, I think, enabled by default on many cheap laptops
because some idiot designer has put the internal microphone right next
to the HD and fan.

All this processing must take time. I don't know whether the signal goes
through the processing hardware or software if processing is switched
out, but I wouldn't be surprised.

Realtek used to have schematics and block diagrams on their site. Not
sure if they still do.

I just tested for it, and am not seeing an echo problem when
recording from microphone input or LineIn input. So not everyone
suffers from this, whatever the problem is. Something the OP
or the youtube ranter are doing with installed software, or
a different product than I'm using, could be doing it. Ventrilo,
TeamSpeak, Skype, some of those things can leave drivers
that run all the time, even when the named products are
not running.

A person could take an empty drive, do an OS test install, and
add elements one at a time, and retest, to figure out when or
why it happened. On Windows 7, you don't even need to type
a license key into the install window - just hit Next and
carry on. On Windows 8, for a test install, you use an
"install only" key to get past the license key prompt.
Test installs will run for enough days, to do the necessary
testing. I use test installs all the time here, when trying
to reproduce a poster's problem. I happened to have a fairly
ripe Win7 install to use in this case, and no echo problem.

Paul

On 05/01/2015 05:30 PM, riserman wrote:
Windows 7 is infamous for latency issues which you can hear demonstrated
on the 'net as windows 7 latency, and which I have personally
experienced. Windows XP did not have this problem.

Does anyone know if Microsoft has eliminated latency in 10?

Thanks.

riserman


Not to ask too stupid a question, but have you test the
problem with VLC yet?

http://www.videolan.org/

VLC plays the most stuff and does it rather well.
(Ugly interface, but it works.)

And have you tried reproducing with a Live CD (as
Stormin' recommended)?

-T

On 5/5/2015 1:33 AM, T wrote:
On 05/01/2015 05:30 PM, riserman wrote:
Windows 7 is infamous for latency issues which you can hear demonstrated
on the 'net as windows 7 latency, and which I have personally
experienced. Windows XP did not have this problem.

Does anyone know if Microsoft has eliminated latency in 10?

Thanks.

riserman

Guys,

I'm not sure you're answering my question. I'm trying to set up for
two-way Skype conversations.

With a microphone and earphones hooked up, there are two problems. One
is an audible delay between what I say and what I hear. Correctly or
incorrectly, I am calling that delay "latency." The second problem is
repetition of the delayed sound something like an echo.

The question is how, set up for a two-way conversation, can I eliminate
both problems?

Thanks,

riserman

riserman wrote:
On 5/5/2015 1:33 AM, T wrote:
On 05/01/2015 05:30 PM, riserman wrote:
Windows 7 is infamous for latency issues which you can hear demonstrated
on the 'net as windows 7 latency, and which I have personally
experienced. Windows XP did not have this problem.

Does anyone know if Microsoft has eliminated latency in 10?

Thanks.

riserman

Guys,

I'm not sure you're answering my question. I'm trying to set up for
two-way Skype conversations.

With a microphone and earphones hooked up, there are two problems. One
is an audible delay between what I say and what I hear. Correctly or
incorrectly, I am calling that delay "latency." The second problem is
repetition of the delayed sound something like an echo.

The question is how, set up for a two-way conversation, can I eliminate
both problems?

Thanks,

riserman


Seeing as I tested Windows 7 SP1 on the test machine,
used a microphone and Audacity (and Sound Recorder),
and no echo was heard, I have to conclude that Skype
has a bad echo suppressor implementation. Or yet
a third piece of software (containing a second echo
suppressor) is present on the computer.

Review *carefully*, all aggressive audio software
you have on your computer. You could even share that
complete list of software with us. If we pop a few
of the names of your aggressive audio software into
Google, who knows what we'll find.

Aggressive audio software is defined as:

1) Software that adds a driver, even when the audio application
is not running. Programs like TeamSpeak and Ventrillo,
those are the class of program that do stuff like this.

Now, if a bunch of UpperFilter or miniport audio drivers
are floating around in the system, who knows what might happen.

You could also check the Skype preferences, and see if there
are any options in there for echo suppression or whatever.

Paul