or Connect
AVS › AVS Forum › Audio › Audio theory, Setup and Chat › Simplified REW Setup and Use (USB Mic & HDMI Connection) Including Measurement Techniques and How To Interpret Graphs
New Posts  All Forums:Forum Nav:

Simplified REW Setup and Use (USB Mic & HDMI Connection) Including Measurement Techniques and How... - Page 89

post #2641 of 9496
Quote:
Originally Posted by basshead81 View Post

any other thoughts jerry?

Sorry, I didn't realize there was an unanswered question. If you are talking about the 30Hz peak, I thought you said you liked it. If you want to tame it, the options are different positions for the sub(s), or DSP of some type. Refresh my memory, are you running Audyssey?

Gear mentioned in this thread:

post #2642 of 9496
Quote:
Originally Posted by AustinJerry View Post

I have workmen in my home today, so I'm looking for things to occupy my time.  I found this interesting read on the HiFiZine site that I thought would be useful to share, authored by Nyal Mellor:  http://www.hifizine.com/2010/09/the-three-acoustical-issues-a-room-correction-product-can%e2%80%99t-actually-correct/.  My thanks to Nyal!

The SBIR has been discussed earlier in this thread by Sanjay. 

In the Early Reflections section, Nyal suggests that all reflections be -10dB relative to the direct sound.  Given this objective, I re-visited my ETC graphs and noted that all my measurements meet this criteria.  No more string technique for Jerry!

And finally, Nyal suggests that with respect to long decay times, a good two-channel listening environment should have a reverberation time of approximately 0.4s.  Mine measures 0.47, slightly higher than optimal.

Interesting. I thought the general recommendation the Acoustic Treatments thread was for attenuation of -15 to -20 dB in gain from the direct signal. Here is my understanding of using ETC for this:

The ETC allows identification of the arrival time and gain of direct versus indirect signals, measured one speaker at a time. Early-arriving, high-gain indirect signals (e.g., 'early specular reflections’) have an impact on intelligibility it they arrive within the Haas fusion zone (+/-20ms of the direct signal): The ear/brain cannot separate the direct and indirect signals, and 'fuses' them together into a 'single auditory event'. This phenomenon is corrected by sufficient attenuation (-15 to -20dB in gain from the direct signal) of the high-gain indirect signals arriving within Haas fusion zone; the ear/brain can then separate the direct and indirect signals.

I could be wrong, but that is what I gleaned from the Acoustic Treatment thread. That said, new research brings new ideas/theories/practices.

Mark
post #2643 of 9496
Quote:
Originally Posted by AustinJerry View Post

Sorry, I didn't realize there was an unanswered question. If you are talking about the 30Hz peak, I thought you said you liked it. If you want to tame it, the options are different positions for the sub(s), or DSP of some type. Refresh my memory, are you running Audyssey?

it does not bother me according to my ears, but i keep wondering what it would sound like if i had that mythical flat response that the forums rave about lol.

no ypao...yamaha 773. i have exhausted about all my sub placement options. my previous graph a couple pages back is what i get when placing them elsewhere. lots of peaks...this seems to be the best looking graph i have come up with minus the one peak. im thinking a mini dsp to tame, but didnt know if you knew of any other tricks? room treatments are not going to pass the WAF lol.
post #2644 of 9496
While standard room treatments can be ugly, don't forget that some absorbers can have artwork silkscreened onto them. Also, very heavy drapes can provide some absorption and bookshelves containing randomly sized books and other objects can act as diffusers.
post #2645 of 9496
Quote:
Originally Posted by AustinJerry View Post

I have workmen in my home today, so I'm looking for things to occupy my time.  I found this interesting read on the HiFiZine site that I thought would be useful to share, authored by Nyal Mellor:  http://www.hifizine.com/2010/09/the-three-acoustical-issues-a-room-correction-product-can%e2%80%99t-actually-correct/.  My thanks to Nyal!

The SBIR has been discussed earlier in this thread by Sanjay. 

In the Early Reflections section, Nyal suggests that all reflections be -10dB relative to the direct sound.  Given this objective, I re-visited my ETC graphs and noted that all my measurements meet this criteria.  No more string technique for Jerry!
The shorthand version quoted there can be quite misleading. I suggest reading Nyal and Jeff's paper instead: http://www.acousticfrontiers.com/static/513e1e34e4b00efcff5b05fd/51523db2e4b05218a1268532/51523db3e4b05218a126876f/1340681100537/acoustic_measurement_standards.pdf. Go to page 5 and notice this:

"Be downto ­10dBby 40ms to prevent breakdown of the precedence effect"

Note the 40 msec criteria. In other words, he is saying make sure you don't hear reflections as distinct echos. This is not a criteria meant for suppression of reflections for the sake of it.

He also makes the important point which was mentioned earlier in this thread that you cannot trust ETC measurements to provide reliable data for this purpose:

"A popular approach issimply to analyze the level ofreflections on an ETC and compare these to the
direct sound,setting a target for the reflections to be 10dB or more less than the direct sound. This
analysis is not sufficient since ETCs are spectrally blind (i.e.they contain no information as to the
spectral content ofthe reflected sound) and the auditory system is very discerning in its requirements
for spectral balance between the direct and reflected sounds in a room."


Simply put, you have no proper measurement to determine what the audible effect of reflections are with ETC. Treatments/furnishings that filter out high frequencies will have disproportional effect on measured ETC.

The right approach here does not use a measurement at all. Simply decide what school you are in relative to reflections being good or bad and treat them using simple means of determining where they occur relative to the surfaces in the room.
Quote:
And finally, Nyal suggests that with respect to long decay times, a good two-channel listening environment should have a reverberation time of approximately 0.4s.  Mine measures 0.47, slightly higher than optimal.
The guidelines here are from 0.4 to 0.6 so 0.47 is certainly fine. Be sure to measure at 500 to 1000 Hz however to get the proper metric here and not unfiltered. Also keep in mind that these are gross measures so I would not worry about two decimal places of accuracy here smile.gif.
post #2646 of 9496
Quote:
Originally Posted by amirm View Post

The shorthand version quoted there can be quite misleading. I suggest reading Nyal and Jeff's paper instead: http://www.acousticfrontiers.com/static/513e1e34e4b00efcff5b05fd/51523db2e4b05218a1268532/51523db3e4b05218a126876f/1340681100537/acoustic_measurement_standards.pdf. Go to page 5 and notice this:

"Be downto ­10dBby 40ms to prevent breakdown of the precedence effect"

Note the 40 msec criteria. In other words, he is saying make sure you don't hear reflections as distinct echos. This is not a criteria meant for suppression of reflections for the sake of it.

He also makes the important point which was mentioned earlier in this thread that you cannot trust ETC measurements to provide reliable data for this purpose:

"A popular approach issimply to analyze the level ofreflections on an ETC and compare these to the
direct sound,setting a target for the reflections to be 10dB or more less than the direct sound. This
analysis is not sufficient since ETCs are spectrally blind (i.e.they contain no information as to the
spectral content ofthe reflected sound) and the auditory system is very discerning in its requirements
for spectral balance between the direct and reflected sounds in a room."


Simply put, you have no proper measurement to determine what the audible effect of reflections are with ETC. Treatments/furnishings that filter out high frequencies will have disproportional effect on measured ETC.

The right approach here does not use a measurement at all. Simply decide what school you are in relative to reflections being good or bad and treat them using simple means of determining where they occur relative to the surfaces in the room.
The guidelines here are from 0.4 to 0.6 so 0.47 is certainly fine. Be sure to measure at 500 to 1000 Hz however to get the proper metric here and not unfiltered. Also keep in mind that these are gross measures so I would not worry about two decimal places of accuracy here smile.gif.

I suppose that is why Dennis Erskine recommends starting the room measurement sequence with near-field measurements (both on-axis and off-axis) of the speakers? That way you know if your speakers have good or bad off-axis response, and they can guide your decision to:

1) Attenuate bad off-axis response with absorption.
2) Diffuse good off-axis response with diffusors.
3) Attenuate and diffuse mediochre off-axis response with a combination device?

Thanks.

Mark
post #2647 of 9496
Quote:
Originally Posted by giomania View Post

I suppose that is why Dennis Erskine recommends starting the room measurement sequence with near-field measurements (both on-axis and off-axis) of the speakers? That way you know if your speakers have good or bad off-axis response, and they can guide your decision to:

1) Attenuate bad off-axis response with absorption.
2) Diffuse good off-axis response with diffusors.
3) Attenuate and diffuse mediochre off-axis response with a combination device?

Thanks.

Mark

Absolutely. Proper acoustic treatment design for a room should generally ALWAYS consider the on and off axis characteristics of the speakers (unless you are building some kind of non-environment control room which aims to absorb all the speaker's reflections in which case it probably doesn't matter what your speakers are doing).
Edited by Nyal Mellor - 4/24/13 at 11:34am
post #2648 of 9496
With respect to the ETC it is a good tool for understanding time of flight for reflections (though you could also just look at the impulse response).

What I have a problem with is that is easy to misapply the ETC since in terms of the absolute SPL values the ETC will look very different depending on which frequency range you are looking at. You can see this yourself by doing a sweep with a top limit of 1kHz and a sweep with a top limit of say 10kHz. The charts will not look the same. The 10kHz chart will look virtually identical to a sweep done out to 20kHz i.e the ETC is dominated by the highest frequencies.

It is also easy to absorb a 'bad reflection' say at -5dB relative to the direct sound by putting a 1" absorber up, thinking that you have now fixed things since the ETC now shows -15dB for the same reflection. All you've done, however, is absorb frequencies above 1kHz, everything else is unaffected, so you have now also timbrally distorted your reflected sound as a result colored what you hear!

post #2649 of 9496
Quote:
Originally Posted by Selden Ball View Post

While standard room treatments can be ugly, don't forget that some absorbers can have artwork silkscreened onto them. Also, very heavy drapes can provide some absorption and bookshelves containing randomly sized books and other objects can act as diffusers.

ahh good suggestions. ill look into some absorbers with artwork, that will make her happy. we want to hang a couple more pictures anyway.
post #2650 of 9496
Quote:
Originally Posted by giomania View Post

Interesting. I thought the general recommendation the Acoustic Treatments thread was for attenuation of -15 to -20 dB in gain from the direct signal. Here is my understanding of using ETC for this:

The ETC allows identification of the arrival time and gain of direct versus indirect signals, measured one speaker at a time. Early-arriving, high-gain indirect signals (e.g., 'early specular reflections’) have an impact on intelligibility it they arrive within the Haas fusion zone (+/-20ms of the direct signal): The ear/brain cannot separate the direct and indirect signals, and 'fuses' them together into a 'single auditory event'. This phenomenon is corrected by sufficient attenuation (-15 to -20dB in gain from the direct signal) of the high-gain indirect signals arriving within Haas fusion zone; the ear/brain can then separate the direct and indirect signals.

I could be wrong, but that is what I gleaned from the Acoustic Treatment thread. That said, new research brings new ideas/theories/practices.

Mark

Quote:
Originally Posted by amirm View Post

The shorthand version quoted there can be quite misleading. I suggest reading Nyal and Jeff's paper instead: http://www.acousticfrontiers.com/static/513e1e34e4b00efcff5b05fd/51523db2e4b05218a1268532/51523db3e4b05218a126876f/1340681100537/acoustic_measurement_standards.pdf. Go to page 5 and notice this:

"Be downto ­10dBby 40ms to prevent breakdown of the precedence effect"

Note the 40 msec criteria. In other words, he is saying make sure you don't hear reflections as distinct echos. This is not a criteria meant for suppression of reflections for the sake of it.

He also makes the important point which was mentioned earlier in this thread that you cannot trust ETC measurements to provide reliable data for this purpose:

"A popular approach issimply to analyze the level ofreflections on an ETC and compare these to the
direct sound,setting a target for the reflections to be 10dB or more less than the direct sound. This
analysis is not sufficient since ETCs are spectrally blind (i.e.they contain no information as to the
spectral content ofthe reflected sound) and the auditory system is very discerning in its requirements
for spectral balance between the direct and reflected sounds in a room."


Simply put, you have no proper measurement to determine what the audible effect of reflections are with ETC. Treatments/furnishings that filter out high frequencies will have disproportional effect on measured ETC.

The right approach here does not use a measurement at all. Simply decide what school you are in relative to reflections being good or bad and treat them using simple means of determining where they occur relative to the surfaces in the room.
The guidelines here are from 0.4 to 0.6 so 0.47 is certainly fine. Be sure to measure at 500 to 1000 Hz however to get the proper metric here and not unfiltered. Also keep in mind that these are gross measures so I would not worry about two decimal places of accuracy here smile.gif.

Quote:
Originally Posted by Nyal Mellor View Post

With respect to the ETC it is a good tool for understanding time of flight for reflections (though you could also just look at the impulse response).

What I have a problem with is that is easy to misapply the ETC since in terms of the absolute SPL values the ETC will look very different depending on which frequency range you are looking at. You can see this yourself by doing a sweep with a top limit of 1kHz and a sweep with a top limit of say 10kHz. The charts will not look the same. The 10kHz chart will look virtually identical to a sweep done out to 20kHz i.e the ETC is dominated by the highest frequencies.

It is also easy to absorb a 'bad reflection' say at -5dB relative to the direct sound by putting a 1" absorber up, thinking that you have now fixed things since the ETC now shows -15dB for the same reflection. All you've done, however, is absorb frequencies above 1kHz, everything else is unaffected, so you have now also timbrally distorted your reflected sound as a result colored what you hear!


I looked closer at a post by Dennis Erskine, which lists a bunch of echo threshold research, and most of it is around 10ms. He also included a graph that I think supports the -10 dB attenuation in gain (from the direct signal) of the high-gain indirect signals arriving within Haas fusion zone. That assumes I am properly interpreting the graph:

Here is the post: http://www.avsforum.com/t/255432/acoustical-treatments-master-thread/8300_100#post_21299669

And for ease of reference, here is the graph that accompanies the post:



Now I am not sure why I had -15 to -20 attenuation in my mind.

Mark
post #2651 of 9496
Quote:
Originally Posted by giomania View Post



I looked closer at a post by Dennis Erskine, which lists a bunch of echo threshold research, and most of it is around 10ms. He also included a graph that I think supports the -10 dB attenuation in gain (from the direct signal) of the high-gain indirect signals arriving within Haas fusion zone. That assumes I am properly interpreting the graph:

Here is the post: http://www.avsforum.com/t/255432/acoustical-treatments-master-thread/8300_100#post_21299669

And for ease of reference, here is the graph that accompanies the post:



Now I am not sure why I had -15 to -20 attenuation in my mind.

Mark

Hi Mark, people often use that graph which I believe originated from Toole's work. The problem is that people correlate detection of reflections with a negative impact on sound quality i.e. all reflections must be bad and we must get them below the threshold of audibility. That is not true as Toole has clearly shown in his research! Now don't get me wrong I am not saying all reflections are beneficial rather I am saying that we can, thanks to the research Toole and others have done, improve upon the generic 'attenuate reflections by -10dB' B or whatever target. Like many things in acoustics it is easy to take the wrong path based on a shallow understanding of the psychoacoustics and acoustic measurements.Two factors that I consider when reviewing those graphs are the direction of the reflection as well as it's spectral content.
post #2652 of 9496
Quote:
Originally Posted by giomania View Post

Early-arriving, high-gain indirect signals (e.g., 'early specular reflections’) have an impact on intelligibility it they arrive within the Haas fusion zone (+/-20ms of the direct signal): The ear/brain cannot separate the direct and indirect signals, and 'fuses' them together into a 'single auditory event'. This phenomenon is corrected by sufficient attenuation (-15 to -20dB in gain from the direct signal) of the high-gain indirect signals arriving within Haas fusion zone; the ear/brain can then separate the direct and indirect signals.
Most of what I've read indicates the opposite. If Haas is to be believed, then early reflections can be as much as 10dB louder than the direct sound before they are perceived as separate signals (see graph below). What are the chances of that happening in your listening room?

Instead, like reading the pits in a CD over and over, early reflections aid intelligibility by giving us copy after copy of the direct sound. As you pointed out, we don't hear early reflections as separate signals, so each copy re-inforces the direct sound we initially imprinted on. A couple decades ago, Benade said "It is quite incorrect to assume that the precedence effect is some sort of masking phenomenon which, by blocking out the later arrivals of the signal, prevents the auditory system from being confused. Quite to the contrary, those arrivals that come in within a reasonable time after the first one actively contribute to our knowledge of the source."

The Haas fusion zone is only a fusion zone when it comes to localization (i.e., we hear the direct sound and its early reflection from one fused location rather than two different directions). For other aspects, like intelligibility, we not only hear early reflections but benefit from them. Everyone from hearing-aid designers to conference room architects will tell you that. The whitepaper from Dirac (the room correction in Datasat pre-pros) begins its section on reflections with "There seems to be consensus in the field that some early reflections actually help make speech more intelligible."

So this phenomenon doesn't need to be "corrected", though some people do prefer early reflections attenuated. But that's a personal preference, not a problem that needs fixing. Besides, attenuating the indirect signal will just make it less audible, not "separate the direct and indirect signals". In order to hear them as separate signals, you would have to make the indirect signal louder than the direct sound.
Quote:
Originally Posted by giomania View Post

I looked closer at a post by Dennis Erskine, which lists a bunch of echo threshold research, and most of it is around 10ms. He also included a graph that I think supports the -10 dB attenuation in gain (from the direct signal) of the high-gain indirect signals arriving within Haas fusion zone. That assumes I am properly interpreting the graph:

Here is the post: http://www.avsforum.com/t/255432/acoustical-treatments-master-thread/8300_100#post_21299669

And for ease of reference, here is the graph that accompanies the post:

The 0dB line on the graph above represents the level of the direct sound. Notice that in the first 30 or so milliseconds, the indirect sound has to be at a higher level (as much as 10dB louder) than the direct sound in order to hear them as separate (echo). Below that threshold, reflections contribute to image shift, what Haas called a "pleasant broadening" of the source (apparently his research subjects found the wider soundstage pleasing).

Also notice that we hear those reflections from the moment they occur, since their lines on the graph extend all the way back to T=0. Some people believe that during some period early reflections are inaudible or suppressed, but the graph you posted shows otherwise.
Quote:
Originally Posted by giomania View Post

Now I am not sure why I had -15 to -20 attenuation in my mind.
Maybe you're remembering that that's where early reflections become effectively inaudible (see graph above).
post #2653 of 9496
Quote:
Originally Posted by sdurani View Post

The Haas fusion zone is only a fusion zone when it comes to localization (i.e., we hear the direct sound and its early reflection from one fused location rather than two different directions). For other aspects, like intelligibility, we not only hear early reflections but benefit from them. Everyone from hearing-aid designers to conference room architects will tell you that.
There are two kinds of speech -- human produced, and mechanically produced. A lot of studies have been conducted on speech intelligibility in classrooms. It was shown that room reflections combine with the direct sound, thus amplifying the desired signal several dB relative to the background noise (students or whatever). Makes perfect sense that this helps intelligibility for classrooms, conference rooms and hearing aids.

But when it comes to speech from speakers, I have not seen anyone argue that intelligibility is enhanced by reflections. First, if the reflections make the audio louder, then that would have been compensated by the level calibration, cancelling that benefit. So, unlike the classroom, the absolute loudness of the speech in a home theater will be the same regardless of the reflections. And so will the intelligibility, according to Toole:
Quote:
Speech intelligibility benefits from some control of reverberation time, so rooms for speech communication, such as classrooms, tend to have short RTs, around 0.5 s. However, there is more to the story. In terms of speech intelligibility in large spaces, it has long been recognized that the early reflections that are a component of the early portion of reverberation are important aids to speech intelligibility. Later reflections contribute nothing useful. In rooms where speech intelligibility is important, therefore, it is more important to pay attention to the reflection pattern in the room than to reverberation itself. Increased early reflection energy has the same effect on speech intelligibility scores as an equal increase in the direct sound energy
post #2654 of 9496
In this context, I don't see much distinction between a human talker and a loudspeaker. Since we're quoting Toole:
Quote:
It is a convenient fact that the directivity of human talkers is not very different from those of conventional cone-and-dome loudspeakers. . . . The consequence of this is that if casual conversation is highly intelligible with one person in the location of the loudspeaker and another in the audience area, then it is probable that loudspeaker reproduction of close-miked vocals will be comparably intelligible.
And much of the vocals being reproduced by loudspeakers at home, especially in movie soundtracks and popular music, is close-miked. So it's not a stretch to imagine that human talkers and loudspeakers behave similarly in this situation.

Besides, wasn't much of the research in intelligibility (Haas, Benade, Bradley, Soulodre, Lochner & Burger, Nakajima & Ando, etc) done using loudspeakers rather than human talkers? If improved intelligibility was observed in the course of doing that research, then why would similar improvements not be heard at home?

Early reflections, as your Toole quote mentions, is for speech intelligibility like turning up the direct sound to get better signal to noise ratio. Except in this case he descrbes "noise" as...
Quote:
Noise, in this context, is everything other than the speech we want to hear. When several people talk at the same time, the noise is speech itself. In music, it is the sound of the band with which a vocalist is singing. In movies, it is everything else in a soundtrack occurring at the same time as the dialogue. For long passages in films and television programming, this is atmosphere-inducing music.
So if improving signal to noise means voices stand out better against soundtracks or songs, then that's a benefit.

BTW, I may be interpreting the bolded part of your quote differently than you, but it seems like he's saying that early reflection energy causes speech to be perceived louder ("has the same effect"), not necessarily measure louder (which, as you said, would be compensated for during calibration).

Even IF we disagree on interpretation or the distinction between human talkers and loudspeakers, do you think that early reflections are detrimental to mechanically produced speech or have a neutral effect?
post #2655 of 9496
carckling.zip 4157k .zip file popping.zip 4246k .zip file
I have been really struggling getting measurements on a couple fronts.

1. I had posted earlier about some REW "headroom issues" and the responses were to either turn up or down the MV but that hasn't seemed to addressed the issue and I'm not sure what to do.

2. Measurements - when performing measurements I get this horrid crackling or popping sound from the sweeps (see attached, you may have to turn up the volume), so I have to "man" the controls diligently in order to mute the volume on the AVR and cancel the sweep => Every time I select "measurement" I don't know if the sweeps will be clean or distorted and am kinda terrified at blowing the speakers (not sure if it would but....). It is random and I will get a undistorted sweep sometimes, but I never can tell and its pretty stressful selecting measure

Any assistance would be greatly appreciated.

System:
- Win7 x64
- Using the HDMI method
- Latest beta REW build
- Followed the PDF config/setup exactly
- EMM-6 (with the cal file downloaded per my serial #) with a MicMate USB adaptor
- Marantz 6006
- Spkrs - KEF IQ90, IQ60C
- Sub - 2x PSA XS15
- Noise floor right around 46db from REW's SPL meter
post #2656 of 9496
Quote:
Originally Posted by sdurani View Post

Instead, like reading the pits in a CD over and over, early reflections aid intelligibility by giving us copy after copy of the direct sound. As you pointed out, we don't hear early reflections as separate signals, so each copy re-inforces the direct sound we initially imprinted on. A couple decades ago, Benade said "It is quite incorrect to assume that the precedence effect is some sort of masking phenomenon which, by blocking out the later arrivals of the signal, prevents the auditory system from being confused. Quite to the contrary, those arrivals that come in within a reasonable time after the first one actively contribute to our knowledge of the source."

 

I just don't get this. Far be it for me to contradict professional acousticians, but I still just don't get it. If the intelligibility is improved by these early reflections, then there must be a problem with the intelligibility in the first place (or it wouldn’t be appreciably improved). So what is the problem? Why, with a high level modern multichannel sound reproduction system, with a centre channel devoted to speech, should there be a problem with intelligibility that requires reflections to 'correct' or 'aid' or 'improve'?

 

In my own system, every time I have added treatments and reduced reflections, centre channel dialogue intelligibility has improved. Why would I need these reflections to help me hear what is coming from the speakers?  Surely what I want to hear is the speakers - and nothing but the speakers?  Adding the treatments aids that, and the dialogue intelligibility improves audibly and without question. Here at least.

 

It seems to me, in my ignorance no doubt, that much of this research is decades old. Well before the time when we had multichannel systems, competent centre speakers and so on. Is such research, conducted often, AIUI, with mono systems or stereo systems, in less than ideal rooms, still to be relied on?

 

I have read a defence of these early reflections based on speech in classrooms. What??  What relevance does that have to our multichannel home theatre systems?  It seems that, in a classroom, the reflections aid intelligibility by making the source appear to be louder.  Well yeah, if I could not actually hear the teacher before, making his voice louder will indeed aid my understanding of what he is saying! 

 

But to make my system louder I don't need to add some reflections. All I need to do is turn up the master volume. 

 

So what am I missing and where am I going wrong? (It's a general question Sanjay, just sparked by your post, so please don't feel that you have to personally respond, although I am always helped by your comments, so if you do, it will be appreciated).

post #2657 of 9496
Quote:
Originally Posted by Roger Dressler View Post

Quote:
Originally Posted by sdurani View Post

The Haas fusion zone is only a fusion zone when it comes to localization (i.e., we hear the direct sound and its early reflection from one fused location rather than two different directions). For other aspects, like intelligibility, we not only hear early reflections but benefit from them. Everyone from hearing-aid designers to conference room architects will tell you that.
There are two kinds of speech -- human produced, and mechanically produced. A lot of studies have been conducted on speech intelligibility in classrooms. It was shown that room reflections combine with the direct sound, thus amplifying the desired signal several dB relative to the background noise (students or whatever). Makes perfect sense that this helps intelligibility for classrooms, conference rooms and hearing aids.

But when it comes to speech from speakers, I have not seen anyone argue that intelligibility is enhanced by reflections. First, if the reflections make the audio louder, then that would have been compensated by the level calibration, cancelling that benefit. So, unlike the classroom, the absolute loudness of the speech in a home theater will be the same regardless of the reflections. And so will the intelligibility, according to Toole:
Quote:
Speech intelligibility benefits from some control of reverberation time, so rooms for speech communication, such as classrooms, tend to have short RTs, around 0.5 s. However, there is more to the story. In terms of speech intelligibility in large spaces, it has long been recognized that the early reflections that are a component of the early portion of reverberation are important aids to speech intelligibility. Later reflections contribute nothing useful. In rooms where speech intelligibility is important, therefore, it is more important to pay attention to the reflection pattern in the room than to reverberation itself. Increased early reflection energy has the same effect on speech intelligibility scores as an equal increase in the direct sound energy

 

 

Phew!  Thanks for that Roger.  I just typed in a lengthy reply to Sanjay asking him why I needed reflections to make dialogue more intelligible in my multichannel HT system. But from your comments here, it seems I am right - I don't!

 

I said to Sanjay "I don't get it" and what I don't 'get' is applying research into speech in classrooms to multichannel HT systems in the home - they just seem to me to be entirely separate things - and as you observe here, and I observed to Sanjay, if I need to make my system louder in order to hear dialogue properly, I don't need to resort to reflections - I need to resort to the volume control.

 

I may be talking out of my fundament, but I can’t personally see any case for where reflections would improve my HT experience. Everything I need, AFAICT, has been 'programmed' into the soundtrack when it was created. And my multichannel system and heavily treated room reproduces it.

 

Sanjay quotes: "The 0dB line on the graph above represents the level of the direct sound. Notice that in the first 30 or so milliseconds, the indirect sound has to be at a higher level (as much as 10dB louder) than the direct sound in order to hear them as separate (echo). Below that threshold, reflections contribute to image shift, what Haas called a "pleasant broadening" of the source (apparently his research subjects found the wider soundstage pleasing)."  My bolding.

 

Again, I just don't get it. A pleasant broadening of the soundstage?  Created by reflections?  Why not create it by moving the L and R speakers slightly further apart?  Or adding 'wide' channels if the former isn’t enough?  

 

From your comments, I think I have come to the conclusion (and please correct me if I am wrong) that research by hearing aid designers and conference room architects has zero relevance to my m/ch HT system, so I can just ignore it all (other than for academic pursuit).

post #2658 of 9496
Quote:
Originally Posted by Ratamacue View Post

carckling.zip 4157k .zip file popping.zip 4246k .zip file
I have been really struggling getting measurements on a couple fronts.

1. I had posted earlier about some REW "headroom issues" and the responses were to either turn up or down the MV but that hasn't seemed to addressed the issue and I'm not sure what to do.

2. Measurements - when performing measurements I get this horrid crackling or popping sound from the sweeps (see attached, you may have to turn up the volume), so I have to "man" the controls diligently in order to mute the volume on the AVR and cancel the sweep => Every time I select "measurement" I don't know if the sweeps will be clean or distorted and am kinda terrified at blowing the speakers (not sure if it would but....). It is random and I will get a undistorted sweep sometimes, but I never can tell and its pretty stressful selecting measure

Any assistance would be greatly appreciated.

You don't provide enough information on your first issue for me to understand what exactly you are experiencing. Can you provide more detail, and perhaps screen shots of any warning or error messages?

Regarding the second issue, I experience this as well. During a typical REW session, it may happen as often as several times. Like you, I want to prevent any potential damage to the speakers, so I have gotten in the habit of keeping the cursor poised over the "Cancel" button. So what causes this? My theory is that it is an interaction between the ASIO beta drivers and something in the laptop (I don't think this is a REW issue, or everyone would be experiencing it). To support this theory, I can take numerous measurements in succession without changing the output source in the REW preferences screen and never get the distorted sound. Then, when I change the output source, e.g. from left speaker to right speaker, the first measurement produces the distorted sound. I fix it by forcing an HDMI handshake, i.e. unplugging and re-inserting the HDMI cable. To be honest, this is yet another annoyance that has detracted from the so-called "simplicity" of using ASIO/HDMI connections. I never had such issues using the legacy REW hardware.

If the issue makes using REW difficult, I would recommend switching to the Java interface and abandoning HDMI/ASIO completely.
post #2659 of 9496
TBH, the discussions on reflections by Nyal, Sanjay, and Roger are interesting, but at the same time frustrating. How is the novice audiophile like myself supposed to evaluate refections in the listening room and determine whether they are "good" or "bad"? And how to apply corrections if needed?

Just when I thought I was making progress, fear and uncertainty is starting to creep back in. And I thought I was enjoying how my system sounds...
post #2660 of 9496
Quote:
Originally Posted by sdurani View Post

Most of what I've read indicates the opposite.

where is the psycho-acoustic research invalidating the haas corollary to the henry precedence effect??
Quote:
Originally Posted by sdurani View Post

Instead, like reading the pits in a CD over and over, early reflections aid intelligibility by giving us copy after copy of the direct sound. As you pointed out, we don't hear early reflections as separate signals, so each copy re-inforces the direct sound we initially imprinted on.

how on earth do multiple signals arriving within the fusion zone (read: time-smearing zone) aid intelligibility? what is "missing" from the direct signal that is so imperative to intelligibility that can only be aided by multiple copies of indirect signals?

and yet there are no issues with intelligibility when wearing headphones - of which there are no indirect signals fusing with the direct signal to apparently "aid intelligibility".

Quote:
Originally Posted by sdurani View Post

The Haas fusion zone is only a fusion zone when it comes to localization (i.e., we hear the direct sound and its early reflection from one fused location rather than two different directions).

this is quite simply false. the best exaggerated case (thus, most obvious) of this smearing is in an auditorium without early reflection control (poor dispersion) or that of mis-aligned PA --- where the gain is more than adequate but the intelligibility of the speaker is nothing but an unintelligible blob.

Quote:
Originally Posted by sdurani View Post

For other aspects, like intelligibility, we not only hear early reflections but benefit from them. Everyone from hearing-aid designers to conference room architects will tell you that. The whitepaper from Dirac (the room correction in Datasat pre-pros) begins its section on reflections with "There seems to be consensus in the field that some early reflections actually help make speech more intelligible."

yes. the conference/speech room architects will most certainly tell you that. this is because the context is different as speech room studies deal with unamplified sources (eg, a human speaking). therefore, the problem with respect to those rooms is the poor signal-to-noise ratio with respect to the direct signal. therefore, intelligibility is "aided" in those circumstances as the "fused" indirect signals increases the perceived gain of the direct signal --- aiding intelligibility by allowing the speech to be "heard".

but in home theaters, we have amplified setups where the gain of the direct signal can be controlled via that of the volume knob.

gain is but one aspect of intelligibility - and most certainly not an issue in residential home theaters (that is, unless your home is built adjacent to a train yard).

you'll notice that all major control/mix/mastering room models fundmentally attenuate the early-arriving high-gain signals that are destructive to intelligibility, localization, and imaging.

all critically-accurate control rooms such as: BBC's CID, RFZ(LEDE), Non-Environment, FTB, Ambechoic (Blackbird C), etc attenuate these destructive indirect signals. if high-gain early reflections aided "intelligibility", then by all means why don't the critically-accurate speech control rooms/mix rooms factor early reflections into their designs to "aid intelligibility"?. instead, they go to great lengths to identify and attenuate.

Blackbird C is one of the most reflection-rich rooms on the planet - but these "early reflections" are -30dB down and thus below the human detection threshold such that the ear-brain is not keying on these signals for localization, imaging, etc ... and thus therefore not "destructive".

we continually see these "unamplified speech room studies" erroneously presented and attempted to be related to home theaters with amplified reproduction setups. context is key!!
post #2661 of 9496
Quote:
Originally Posted by AustinJerry View Post

Regarding the second issue, I experience this as well. During a typical REW session, it may happen as often as several times. Like you, I want to prevent any potential damage to the speakers, so I have gotten in the habit of keeping the cursor poised over the "Cancel" button. So what causes this? My theory is that it is an interaction between the ASIO beta drivers and something in the laptop (I don't think this is a REW issue, or everyone would be experiencing it). To support this theory, I can take numerous measurements in succession without changing the output source in the REW preferences screen and never get the distorted sound. Then, when I change the output source, e.g. from left speaker to right speaker, the first measurement produces the distorted sound. I fix it by forcing an HDMI handshake, i.e. unplugging and re-inserting the HDMI cable. 

Similar here. Doesn’t happen very often and seems to be unpredictable. I assumed it was just happening here. Like you, restarting REW or replugging HDMI and mic seems to fix it. 

post #2662 of 9496
Quote:
Originally Posted by AustinJerry View Post

TBH, the discussions on reflections by Nyal, Sanjay, and Roger are interesting, but at the same time frustrating. How is the novice audiophile like myself supposed to evaluate refections in the listening room and determine whether they are "good" or "bad"? And how to apply corrections if needed?

Just when I thought I was making progress, fear and uncertainty is starting to creep back in. And I thought I was enjoying how my system sounds...

 

I sympathise with your position Jerry. Personally, I regard all reflections as 'bad'. I cannot see any way that they can help the SQ in a modern m/ch HT. They may have a positive role to play in a 2 ch stereo system - IDK or care about that. Neither do I care if my room is 'dead'. All the ambience etc I need is recorded into the soundtrack and when it plays back, I hear the speakers and they reproduce whatever was recorded, which is what I want. I have never had such good sound at home as I currently have (biggest improvement came from room treatments) so I don't worry too much about it any more. Like you, I will never understand all the nuances of acoustics and I think it is easy to be misled by information that is not 100% transferable to our own realities. 

 

Bottom line: if you are enjoying it, just continue to enjoy it.

 

EDIT:  Just read Localhost's post above. He says with far greater authority what I was trying to say to Roger and Sanjay and FWIW I agree entirely with the content of his post. We just don't need these reflections in our systems to aid intelligibility. Good point made too about when listening with headphones where there are no intelligibility issues and also no reflections!  If ever I move my HT system into a classroom, maybe I will think differently ;)


Edited by kbarnes701 - 4/26/13 at 7:04am
post #2663 of 9496
Quote:
Originally Posted by sdurani View Post

So if improving signal to noise means voices stand out better against soundtracks or songs, then that's a benefit.

the quality of the "mix" of the source material is another factor entirely to the discussion.
post #2664 of 9496
Quote:
Originally Posted by localhost127 View Post

Quote:
Originally Posted by sdurani View Post

So if improving signal to noise means voices stand out better against soundtracks or songs, then that's a benefit.

the quality of the "mix" of the source material is another factor entirely to the discussion.

 

Thanks for pointing this out. I am reluctant to enter into discussions with people who are far more knowledgeable than I am but this was another thing I 'didn't get' when I read Sanjay's post earlier. WRT to signal to noise, well, there isn't really much 'noise' in my HT other than that generated by the speakers, so I can discount any alleged benefits of early reflections in that regard. And as for making dialogue stand out better against background noises on movie tracks, well, I assume the mixer will have taken care of that for me. So if there is a problem hearing the dialogue properly then it has to be a problem with my room or system. I have to say that since treating the room heavily I have zero problems with dialogue intelligibility. And of course, if the mixer did a bad job and the background sounds in the movie overwhelm the dialogue, then there isn't much I can do about that anyway. And anything I did try to do would be counterproductive because as soon as I came back to a well-mixed track, my 'adjustments' would contaminate that track and certainly not improve it. I guess I really do 'just not get' all this stuff about early reflections improving intelligibility. Either that or I am (improbably) right and they don't ;)  Well not in a m/ch HT environment anyway. 

post #2665 of 9496
Quote:
Originally Posted by kbarnes701 View Post

I regard all reflections as 'bad'. I cannot see any way that they can help the SQ in a modern m/ch HT. They may have a positive role to play in a 2 ch stereo system -

there are multiple "perspectives" within an acoustical space of which to view from.
take Non-Environment control rooms for example. a NE response (in the time-domain/ETC measured at the listening position) will show the direct signal and then drop off to the ambient noise floor. the speaker-to-room response is effectively anechoic - no indirect energy from the speakers (via the room / reflections) is allowed to impede the listening position. the direct signal is all that is heard/processed/measured at the listening position (for "accuracy").

now, some people will assume this "effectively anechoic" response as measured at the listening position must mean that the room itself is dead/anechoic - which can be quite uncomfortable to be in! but this isn't true at all as the front wall and floor are reflective.

so, the speaker "sees" an anechoic room (speaker-room response), but the human/listener does not. when the human speaks, moves about, etc ... there are indirect room reflections. the difference is that in this case the human is the "source" and "receiver" - and this "response" is NOT anechoic as energy reflects from the floor and front wall. but when the speaker is the "source" and the listening position (mic) is the "receiver", then the response is anechoic.

so you can indeed construct the room such that the speaker-to-listener response is anechoic and accurate ... while the room itself does not "sound dead" to the listener. it's all about perspective wink.gif

it's not just the time-arrival but the gain of the energy. you'll find Binary Amplitude Diffusers (RPG BAD, for example) common in many home theaters. simply using absorption to attenuate the high-gain, sparse, early arriving signals can quickly lead to a dead room (from the human's perspective) - especially when confronted with many rows/seating positions (thus, many "reflection points"). using BAD can provide mid-LF absorption while also offering HF spatial dispersion. spreading the energy out in many directions (diffusing) by nature lowers the gain of the energy that reflects specularly to the listening position. thus, you can have some energy/reflection but it is low enough in gain as to not be destructive. it also provides a sense of "reflected energy" to the humans as they talk and move about the room and restrict the "deadness".




in a 2ch there are multiple approaches. if one is concerned with accuracy with respect to the direct signal (and minimizing the masking/skewing effect their room imparts on what is heard/processed at the listening position), then the high-gain (sparse) indirect reflections within the haas interval are indeed attenuated below human detection threshold such that they are not destructive to intelligibility, localization, and imaging.

but that doesnt mean they must be eliminated (far too commonly misinterpreted here).

other 2ch time-domain responses do not "absorb" these early sidewall reflections but instead simply delay them in time by redirecting them to the rear of the room where they can be diffused and re-introduced to the listening position (later in time, outside of the haas interval!) as a laterally-arriving, exponentially decaying diffused sound-field. this is accomplished via 1dimensional reflection phase grating diffusers installed on the rear wall/rear side-walls to provide this "passive" surround sound / lateral returns for envelopment --- with the wells oriented vertically to provide spatial dispersion in the horizontal/lateral plane.

eliminating the sparse reflections also greatly reduces the 3d spatial polar lobing which translates into the frequency response as the "comb-filter" interference pattern. the sparse comb-filtering is detrimental - but we can use diffusers to "break-up" these sparse reflections into MANY reflections dispersed spatially and temporally in time ... which makes the comb-filtering more complex and dense - greatly minimizing the frequency response anomalies while providing spaciousness (the "well-mixing" of the reflections/acoustical energy via the diffusers - of which happens naturally in larger, more "spacious" rooms).

again, care must be taken not to lump all reflections together. gain, time-arrival, spectral content, ingress angle, sparse/diffused, etc are all characteristics of acoustical energy. we don't refer to lakes, glaciers, icecubes, rivers, oceans, mist, fog, steam, rain as simply "water" ... yet we often see all forms of acoustical energy lumped together as simply "reflections". some 2ch models go to great lengths to attenuate these destructive early arriving signals while at the same time striving to preserve the acoustical energy such that it can be reintroduced to the listening position at a later time.
Edited by localhost127 - 4/26/13 at 8:00am
post #2666 of 9496
Quote:
Originally Posted by Roger Dressler View Post

There are two kinds of speech -- human produced, and mechanically produced. A lot of studies have been conducted on speech intelligibility in classrooms. It was shown that room reflections combine with the direct sound, thus amplifying the desired signal several dB relative to the background noise (students or whatever). Makes perfect sense that this helps intelligibility for classrooms, conference rooms and hearing aids.

But when it comes to speech from speakers, I have not seen anyone argue that intelligibility is enhanced by reflections. First, if the reflections make the audio louder, then that would have been compensated by the level calibration, cancelling that benefit. So, unlike the classroom, the absolute loudness of the speech in a home theater will be the same regardless of the reflections. And so will the intelligibility, according to Toole:
As Sanjay mentioned, almost all speech research is done with speakers. I have quoted many and can dig them up if needed.

On the Toole quote, it is critical to make distinctions in reflections:

1. Those that occur below the timing of a typical vowel in English. That is about 0.08 seconds. Once you go beyond this, the reflections start to cause temporal masking and hence degrade speech. This is what the Toole quote meant that you post. These are called late reflections and indeed are enemy of good speech although for other content, they continue to add a pleasing effect. Hence the typical recommendation of keeping late reflections (measured using RT60) in the 0.4 to 0.6 second range as a balance between speech quality and enjoyment factor.

2. The early reflections below the vowel time do increase speech intelligibility as mentioned since they increase the sound power. There simply is more energy in the room once you allow the reflections to exist and that improves our ability to hear it. Here is a much more useful version of the graph being posted for this domain (from Dr. Toole's book created from composite research):

i-wKXwcMt-L.png

First we see that the content type is speech. For other types such as clicks and noises as Dennis' post mentioned, the thresholds change. That aside for now, we see that the first reflections as noted by the dots are safely in the fusion range (i.e. not hearing them as echos as Sanjay mentioned) for typical listening rooms. And hence, there is nothing to "fix."

Back to Dennis' post, how or why we hear early reflections this way is not fully understood. Some of it is due to how the auditory system works, but some is based on the cognitive portion of our brain that analyzes and thinks about what is being heard. There are observations for example that what you have heard in the past, may change what the thresholds are in the above graph!

So to summarize, you don't want to have an overly live room for the reasons that Roger mentions. Measure mid-frequency RT60 using REW and target 0.4 to .6 seconds. For early reflections, you should buy speakers that either have good off-axis response or limited amount. Once there, the default choice should be to leave side reflections be. Front, floor and ceiling and back can be absorbed if your room is not too dead per the RT60. If you have less than well designed speakers, then the research still says that people tend to prefer hearing the reflections than not although people argue that point smile.gif. It seems to me though that it makes little sense to spend many hours measuring and futzing with your room if your speakers are not well designed. Replace them first, and then worry about the room.
post #2667 of 9496
Quote:
Originally Posted by kbarnes701 View Post

Far be it for me to contradict professional acousticians...
Since I'm not a professional, just a hobbiest like you, don't hesitate to contradict anything I post.
Quote:
Originally Posted by kbarnes701 View Post

If the intelligibility is improved by these early reflections, then there must be a problem with the intelligibility in the first place (or it wouldn’t be appreciably improved).
Unless you believe absolute perfection exists, there is always room for improvement. Titles pop up (e.g., Star Trek 2009) that get more than usual complaints about dialogue being hard to hear. In those situations, making it easier to pick out voices can be helpful.
Quote:
Originally Posted by kbarnes701 View Post

But to make my system louder I don't need to add some reflections. All I need to do is turn up the master volume.
But that makes everything (signal and noise) louder, which isn't that helpful when you're trying make voices easier to hear against a noisy soundtrack or loud song. One alternative is to raise the volume of just the centre channel, but then you've changed the calibrated balance of the L/C/R speakers.

BTW, it's not about the need to "add" reflections, since they are already there as a naturally occuring aspect of the room. Where you and I disagree is the need to remove them.
post #2668 of 9496
Quote:
Originally Posted by giomania View Post

Interesting. I thought the general recommendation the Acoustic Treatments thread was for attenuation of -15 to -20 dB in gain from the direct signal. Here is my understanding of using ETC for this:

The ETC allows identification of the arrival time and gain of direct versus indirect signals, measured one speaker at a time. Early-arriving, high-gain indirect signals (e.g., 'early specular reflections’) have an impact on intelligibility it they arrive within the Haas fusion zone (+/-20ms of the direct signal): The ear/brain cannot separate the direct and indirect signals, and 'fuses' them together into a 'single auditory event'. This phenomenon is corrected by sufficient attenuation (-15 to -20dB in gain from the direct signal) of the high-gain indirect signals arriving within Haas fusion zone; the ear/brain can then separate the direct and indirect signals.

I could be wrong, but that is what I gleaned from the Acoustic Treatment thread. That said, new research brings new ideas/theories/practices.

Mark

Quote:
Originally Posted by giomania View Post



Now I am not sure why I had -15 to -20 attenuation in my mind.

Mark

localhost127's posts here jogged my memory enough that I was able to find his post where I (incorrectly, I think) interpreted the -15 to -20 dB attenuation recommendation. In the post he prefaces the statement with "keeping things simplified", so once again, it probably depends.

http://www.avsforum.com/t/255432/acoustical-treatments-master-thread/7900_100#post_20914820

That said, this is a great discussion. I know it is off topic, but Jerry started it! tongue.gif

Mark
post #2669 of 9496
Quote:
Originally Posted by AustinJerry View Post

How is the novice audiophile like myself supposed to evaluate refections in the listening room and determine whether they are "good" or "bad"?
You can start by worrying less about what's "good" or "bad" and instead tune your room for what you like, assuming you're willing to use personal preference as goal (not everyone is comfortable with that notion, as you've no doubt read in the reference vs preference discussions in the Audyssey thread).
Quote:
Originally Posted by AustinJerry View Post

And how to apply corrections if needed?
Use your ears. I don't mean that to sound glib, but that's how I arrived at it. Following 'conventional wisdom', I tried absorbers and diffusors at the early reflection points olong my sides, only to keep returning to bare walls. Couldn't figure out why my tastes were so unusual until I read the research on reflections and found out that: a) my preferences were more common than not, and b) many of the problems I was warned about turned out either not to be problems (comb filtering) or things I liked (image shift - I loves me a big, wide soundstage). Eventually I found that I preferred my absorbers on the front wall rather than the side walls. But the only way to know that was by trying and listening.
post #2670 of 9496
Quote:
Originally Posted by giomania View Post

interpreted the -15 to -20 dB attenuation recommendation.
Mark

yes, to be below human detection threshold (-30dB ideally) such that the ear-brain is not keying on that signal for localization and imaging.

the point is, if you choose to attenuate a particular indirect specular signal, then a broadband treatment must be utilized (absorber, diffuser, angled-reflector, etc).
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Audio theory, Setup and Chat

Gear mentioned in this thread:

AVS › AVS Forum › Audio › Audio theory, Setup and Chat › Simplified REW Setup and Use (USB Mic & HDMI Connection) Including Measurement Techniques and How To Interpret Graphs