Soundstage Width cannot extend beyond speakers

[STC]

7 hours ago, esldude said:

I posted some sample music one file of which should image outside the edges of your speakers.  I've not kept up with this thread nor do I intend to, but thought it might be of interest as it is on topic for this thread title.  So visit the thread link above. 

 

 

 

There is problem with using Larry's workaround test sample for this purpose. The recording was recorded in a church where the reverberation will be high. Reverberation and reflection will have out of phase information which can shift the image.  Ideally, such recordings should be done in free field so that only the exact position of source is captured without any reflection.

 

I still have your cricket recording and if I remember correctly there was a vehicle traveling at far right or left. Do you have any idea what would be angle of the cricket, the air cond and the traveling vehicle? 

 

[STC]

16 minutes ago, pkane2001 said:

 

I think there are at least a few problems with your argument here.

 

For one thing, it's not clear to me that precedence (the effect of discarding reflections) is at all useful when talking about time difference of the sound between the ears. As I understand it, a ear must receive at least the first wavefront before it can cancel the subsequent, reflected one and ignore it. Why do you think that a time delay between the ears will do the same? In the case of ITD, the difference in the edge/fall of a sine wave arrival between the two years is what's used to localize the sound. Does echo cancellation apply to two sounds captured by different ears? I'm not so sure.

 

My guess is that you might be talking about the 'summing' effect, where echos coming into the same ear in 1ms or less appear to add up to a single sound with a greater amplitude. This again, seems to be a completely different mechanism than ITD. ITD does not process two sounds in one ear, it process the differences in time between two sounds in two different ears. What's more, this summing effect applies to ILD (it changes the perception of sound level in one ear). It has no effect on ITD -- the timing difference of the signal between the two ears.

 

Precedence effect is about the arrival of the second sound ( first wavefront). It is not necessarily must be reflections. If you see Bluaert papers the whole experiment about precedence effect was described with experiment using two speakers in anechoic room. 

 

You have to understand this before moving on to next stage. 

[STC]

2 minutes ago, esldude said:

Working from memory the primary near source of sound was 10 degrees or so left.  The vehicle would have been far left 80 degrees or so.  The AC would have been behind and between houses with most of that sound multiple reflections.  I think there was another AC perhaps more noticeable which would have been 40-50 degrees right.  

 

The image only shifted so much after I processed it.  And from experience with blumlein miking, if you get a musician over to one side and they get picked up on the rear out phase part of one channel and the front in phase part of the other the odd phasing can put them outside the edges much the same.  Just as is happening with the reverb in a church. 

 

The processing I did to the recording was to duplicate channels.  Swap right for left in the duplicate.  Invert phase of the duplicate.  Drop the level by 6 db.  Then mix it back together. 

 

Rereading your earlier post on the recording of yours, I suspect you have some phase issues with the mics arrangement.  Image shouldn't shift. If someone stands 30 degrees to the left he should be localize there in the recording. 

 

Thanks for the info about the crickets recording. I just made one in a dry environment. Will post soon and see if this sound goes outside your speakers boundary. 

[STC]

24 minutes ago, pkane2001 said:

 

STC, I have no idea what and where you're reading. If you want to continue a discussion, then at least provide a quote from this book that explains your position. Telling me to read a book written in 1973 to understand what you're saying is a very unconvincing argument.

 

Here's a more recent book quote for you, published in 2009:

 

So what does precedence have to do with ITD?

 

Ok. This is going to be more harder  to explain. Even Toole put it under reflection. Anyway, Haas’s experiment was referred by Toole and only two loudspeakers were used. One was the reference and the other was for the delayed sound.

 

What Haas found was from 0 to 0.6 to 1ms, the image shifts towards the second sound depending on the intensity of the second sound. This is the basic for stereophonic.  From 1ms to 20ms , the image did not move but added fullness(sic) to the sound. Forgotten the correct terminology . 

 

I think the confusion here is you are taking the word reflection literally.  A reflection is just another sound of the original sound. For me it is easier to understand because, I do not have reflection in my room but only speakers to produce the reflected sound. So I could experiment this easily even with a single click. 

 

I am unable to give you the reference now to show what happens when the third sound arrives but it will not be heard as separate sound. 

 

I will explain what ITD got to do with this shortly. Need to look for the charts. 

[STC]

7 minutes ago, esldude said:

I'm not sure which recording you are referring to in this instance. If talking about when I've used blumlein and had issues you've misunderstood where they were. 

 

My bad. I was referring to another recording where you mentioned ghosting effect. Probably George. My apologies. 

[STC]

2 hours ago, pkane2001 said:

So what does precedence have to do with ITD

 

Going back to the chart. 

 

 

 Previously, it was calculated that there  was a 410microseconds delay between the left and right ears for the violin signal at 120 degrees. This time difference has already been encoded in the recording. 

 

This difference is represented by L3 and R3. Now, going back to Haas experiment about first wavefront, the second signal L3a will arrive at the right ear about 250 microseconds for 60 degrees. Our brain will now localize the sound to come from 60 degress. 

 

However, the original delay of 410 microseconds ( R3) is yet to arrive at the right ear. R3 will arrive 160microseconds  ( 410 - 250) after L3a. 

 

Now your ears hear three signals delayed by 250 and 160 microseconds. Then there will be another delayed signal R3a that will arrive 250us after the arrival R3.  The difference between L3 and R3a is 660us. 

 

From Haas experiment, and your reference of precedence effect what will happen to R3 and R3a delayed signals that arrive at the ears. Will the image shift or will it superimpose on the original perceived image of the first delayed sound L3a or will it be treated as early reflections?

 

You can use the same principle by moving the speakers closer to the side wall so that the image could shift to the outer of speakers boundary. But the distance to the wall should be very close so that the delay does not exceed 0.6 to 1ms. 

[STC]

14 minutes ago, pkane2001 said:

 

Reading (and interpreting) research, adding up time delays up to 1ms does not cause the precedence effect, but instead causes the summation effect. This simply shifts the apparent sound source in the direction of the ear with the shortest delay. So, the violin will shift to the left, if I understand this correctly.

 

By how many degrees based on the speakers location in the diagram? Can the shift reflect a 410microseconds interaural time difference location? It can't.

[STC]

36 minutes ago, pkane2001 said:

 

Why not? I'm sorry, but I'm missing something in your argument. If a 410usec delay is added to whatever other ITD is produced by the speakers (which is the same from the left speaker and the right speaker), then the sound will shift exactly by a distance represented by 410usec delay. 

 

Didn't you also asked me to read about first wavefront principle?  It can be answered using that.

 

Based on the principle,  the precedence effect takes place after 1ms (sic). Whatever first "reflected" sound that arrives you ear within 1ms will shift the image.  In the diagram, the second sound that arrive the ear is L3a. That will be second loudest and most identical to L3. That sound arrives after 250μs. This will be localized by our brain. That location will be fixed. Now, R3 and R3a will arrive much later. The total time difference between TWO successive signals in the following order L3a, R3 and R3a never reached the real value of 410μs of the original time difference for the violin location. The longest ITD that the brain could receive was 250μs only. Since there is no clear ITD value of 410μs with the speakers reproduction, could you help me to understand how is it possible that the brain now can localize the sound outside the speakers boundary?

 

 

[STC]

6 minutes ago, pkane2001 said:

 

Don't recall asking you about first wavefront. But again, under 1ms, there is no perception of 'first', 'second' or 'third' sound. The sounds coming in with different time delays within 1ms all add up to one perceived source (called the summation effect).

 

 

Ok. Under summation effect where  do you think the violin will be located? It is easy to calculate and let’s make it simple by agree the spread of sound rounded to 340m/s. Can you calculate where the phantom image of the violin will be produced by the speakers?  

[STC]

24 minutes ago, pkane2001 said:

And yes, I used phase in the frequency domain to make these adjustments, not time:

 

You labelled the files with time not phase despite you want to prove IPD is ITD. Why is that despite saying that they are both the same,  you are still bringing in time to illustrate this point. Just stick to phase alone. What is the phase angle for each file?

 

This is about human localization process not to prove that time is phase which is mathematically correct. Any school kids should able to confirm that. I am using ITD to illustrate human localization process. If you want to explain IPD to prove that it can be done use my chart and give the angles and discuss that entirely in phase domain without bringing time. I never used phase in my diagrams. Why it is so complicated to make a point which is so obvious?

[STC]

4 minutes ago, esldude said:

For more fun, slow the file down to 44.1 khz and listen.  I do mean slow down not resample.  In Audacity you can change the file rate without changing samples.  Or alternatively slow the file down to 23% of its normal speed.  Then listen again. 

 

Also with the original file try listening centered, over to the left in line with the left speaker and ditto for the right. 

 

Amazing! try listening at 0.16x speed. Nice echo.

[STC]

4 hours ago, pkane2001 said:

 

Per @jabbr suggestion: here is a test track with a recorded bell and voice. Both are centered on the original track (delay of 0) but recorded at various heights. I've delayed the right channel by various amounts relative to the left.

 

And yes, I used phase in the frequency domain to make these adjustments, not time:

 

IPDTestFiles

 

I'll keep this up for only a short time, so please download and try if you're interested. This has 0, 200, 400, 600, and 1000 microseconds delay in the right channel.

 

 

4 hours ago, STC said:

 

You labelled the files with time not phase despite you want to prove IPD is ITD. Why is that despite saying that they are both the same,  you are still bringing in time to illustrate this point. Just stick to phase alone. What is the phase angle for each file?

 

This is about human localization process not to prove that time is phase which is mathematically correct. Any school kids should able to confirm that. I am using ITD to illustrate human localization process. If you want to explain IPD to prove that it can be done use my chart and give the angles and discuss that entirely in phase domain without bringing time. I never used phase in my diagrams. Why it is so complicated to make a point which is so obvious?

 

 

 

[STC]

There are many recordings could do that. This is about true stereo recordings that work mostly on level difference alone irrespective of the frequency based on stereophonic principle.