[sander99]

I also was in HPA -> HPB comparison mode, and I think it links the azimuth.

Could the following be the case (easy to check this I think):
-When switching on the HPA -> HPB comparison mode the linking of the A and B head tracking is activated.
-When switching off the HPA -> HPB comparison mode the linking should be deactivated but Smyth forgot to program this step.
-Maybe after switching off and on or "power cycling" the linking is deactivated.

 

[Eich1eeF]

There is a setting Dolby Night mode (and DTS Night mode), page 55 of the manual. You could check if they are on or off and if changing them makes a difference. But it would be a very, very strange implementation if these simply changed the (user-) volume setting! I'd sooner believe it is a bug than that it is the night mode, but please do check.

Edit: but it would not be random if it's the night mode, that's why I asked before if repeating the same sequence of scenes gives the same (relative to be exact) changes.

I assumed that night mode should be entirely managed by the APM-89L, and only be turned on/off by the A16's microcontroller. The first thing I thought about was dialnorm https://en.wikipedia.org/wiki/Dialnorm, but that also should not visibly change the volume setting. Additionally, I thought dialnorm would be set once per movie, not per scene.

 

[Got the Shakes]

When I first got the A16 back in August, I was a bit underwhelmed by the preloaded PRIRs especially when compared to the 5.1 PRIR I captured myself. For the heck of it I was messing around with with some Atmos content and decided to make an Atmos listening room from the Surrey Uni speakers in both a 7.1.4 and 9.1.6 setup. I don't know if it is the fact that my brain has been trained at this point or what but I was actually quite impressed by the result of Atmos content with the preloaded PRIR. I'd still like to capture a real system, but what I heard yesterday and today was good enough for me to continue to use it for my Atmos content.

 

[Erik Garci]

To something else: I had an idea, since A16 PRIRs can not be used with the A8, can't you record an A8 PRIR through the A16?

Yes, it's possible, and I think it would sound the same.

Problem would be the delay that adds up, but at least with line input delay of the A16 should not be as huge as with Dolby for example.

I think the A8 would let you reduce whatever delay was measured in the PRIR.

If you don't want to risk the mic ins of the A8 you have to use the A8 mics in your ears and headphones connected to the A16 with the correct HPEQ. But this would of course be another step that could degrade the signal.

For what it's worth, I heard no difference when I compared these on my A8:
1. Stereo PRIR and HPEQ measured by line out directly to mic in (no actual mic).
2. Stereo PRIR and HPEQ measured by line out to headphones to actual mic.
3. Stereo pass-through (no PRIR or HPEQ).

 

[audiohobbit]

I was a bit underwhelmed by the preloaded PRIRs

Did you create a manLOUD HPEQ specifically for this room?
The manLOUD procedure helps a lot to personalize a PRIR that is not made with your ears.
Seems that people still don't get this, and why this is needed and how it's done, although it's relatively good documented in the video.

I only use the Surrey room with manLOUD HPEQs with my Audeze LCD2 and my Hifiman In Ears.
The result with the LCD2 is quite good, I'd say 80 percent of a persoanlized PRIR, with the In Ears maybe 70%.

In the manLOUD process you have to select the specific PRIRyou want to make the manLOUD HPEQ for and you have to select loudspeakers fromt that PRIR, through which you listen to the band limited noise. You have to select Equal Loudness 80 which should be played at 80 dB (you have to guess this, 80 dB noise is quite loud already). Each noise band has to play at the same volume, like it would when listening to real loudspeaker with a linear frequency response. You have to bring each noise subband to the same volume as the others by listening and comparing, not totally easy, but feasible.
From the PRIR you can select the L+R speakers, the C speaker or L+C+R.
So you're manually correct your HRTF for that or these speakers. Normally the real HRTF for each speaker (direction) would be different, so it's a compromise and the front speakers are the most important.
For the manLOUD process the Realiser takes only the direct sound part of the impulse response, not the reverb. Therefore you don't hear the loudspeaker out of your head while doing the HPEQ.

This is a first stage in the direction of personalisation of a different PRIR (or BRIR). On the exchange website they will do this automatically by comparing the part of the impuls response without the reverb to a PRIR that you upload and this has to be personalised to you and speaker positions has to correspond to those in the PRIR you want to personalise.
The HRTF that you correct only for the fronts with the manLOUD process will be corrected automatically for every speaker direction in the automatic process, so the result will be quite good I think and satisfy many people already.

But even with the manLOUD process many people could already be happy with the factory PRIRs, but it's absolutely mandatory to do the manLOUD process with the PRIR. And since you have 3 possibilities of selecting loudspeakers from the PRIR you should try all of them if the first try is not satisfactory. This happened with my In Ears at first. For my LCD2 I only tried L+R and am quite satisfied (haven't tried the other options though, maybe they're even better?). For the In Ears I also tried L+R at first but the result was not good. Then I did it again with C and L+C+R. Both delivered similar and far better results than only L+R. Still a little bit harsh with voices sometimes but OK.

If I find some time I may write a tutorial for this, because I think it's important.

Listening to the factory PRIRs with an autoHPEQ or flat HPEQ is of no use and will deliver horrible results for the majority of people I think.
With the manLOUD procedure I had to reduce some bands 10-15 dB for example! So you can imagine how it sounds when 1 kHz or so will be 15 dB too loud...

With the manLOUD process the Smyths already implemented a very cool and handy feature which I think many people don't yet understand completely.

 

[romainhc]

At one point my volume went from 86 to 74.

Do you use the "tri-vol" option ? Appendix 25

Because you can store 3 volume levels by preset. You recall those levels with the rocker buttons on the remote.

 

[audiohobbit]

On the rotary buttons, not the remote.

Ok, understood.

I also was in HPA -> HPB comparison mode, and I think it links the azimuth.

Yes it definitely does. The manuals says:

The HPB→A mode, while enabled, also correctly switches the headtracking input signal from User A to User B.

"while enabled" could mean as long as this option is switched on in the system menu, not only when actually rerouting B to A via arrow key in speaker map mode.
I tried this yesterday and it seems that it is so. So you have to switch that off in the system menu ( It's now in Misc Settings with the new FW). It's not meant ot be on all the time I think.
So, no bug.


@erik:
Would be very cool if this works. Then you could record a PRIR with the A16 and "convert" this to the A8, without the need to record the same PRIR in the same room with the A8 also.

Stereo PRIR and HPEQ measured by line out directly to mic

Did you just connect the line out to the mic in, without any adapters (impedance, voltage)?? That's brave...
The mic in would need a very low voltage. So the level of the line out has to be set very low, which deteriorates S/N ratio.

 

[Got the Shakes]

Did you create a manLOUD HPEQ specifically for this room?
The manLOUD procedure helps a lot to personalize a PRIR that is not made with your ears.
Seems that people still don't get this, and why this is needed and how it's done, although it's relatively good documented in the video.

I only use the Surrey room with manLOUD HPEQs with my Audeze LCD2 and my Hifiman In Ears.
The result with the LCD2 is quite good, I'd say 80 percent of a persoanlized PRIR, with the In Ears maybe 70%.

In the manLOUD process you have to select the specific PRIRyou want to make the manLOUD HPEQ for and you have to select loudspeakers fromt that PRIR, through which you listen to the band limited noise. You have to select Equal Loudness 80 which should be played at 80 dB (you have to guess this, 80 dB noise is quite loud already). Each noise band has to play at the same volume, like it would when listening to real loudspeaker with a linear frequency response. You have to bring each noise subband to the same volume as the others by listening and comparing, not totally easy, but feasible.
From the PRIR you can select the L+R speakers, the C speaker or L+C+R.
So you're manually correct your HRTF for that or these speakers. Normally the real HRTF for each speaker (direction) would be different, so it's a compromise and the front speakers are the most important.
For the manLOUD process the Realiser takes only the direct sound part of the impulse response, not the reverb. Therefore you don't hear the loudspeaker out of your head while doing the HPEQ.

This is a first stage in the direction of personalisation of a different PRIR (or BRIR). On the exchange website they will do this automatically by comparing the part of the impuls response without the reverb to a PRIR that you upload and this has to be personalised to you and speaker positions has to correspond to those in the PRIR you want to personalise.
The HRTF that you correct only for the fronts with the manLOUD process will be corrected automatically for every speaker direction in the automatic process, so the result will be quite good I think and satisfy many people already.

But even with the manLOUD process many people could already be happy with the factory PRIRs, but it's absolutely mandatory to do the manLOUD process with the PRIR. And since you have 3 possibilities of selecting loudspeakers from the PRIR you should try all of them if the first try is not satisfactory. This happened with my In Ears at first. For my LCD2 I only tried L+R and am quite satisfied (haven't tried the other options though, maybe they're even better?). For the In Ears I also tried L+R at first but the result was not good. Then I did it again with C and L+C+R. Both delivered similar and far better results than only L+R. Still a little bit harsh with voices sometimes but OK.

If I find some time I may write a tutorial for this, because I think it's important.

Listening to the factory PRIRs with an autoHPEQ or flat HPEQ is of no use and will deliver horrible results for the majority of people I think.
With the manLOUD procedure I had to reduce some bands 10-15 dB for example! So you can imagine how it sounds when 1 kHz or so will be 15 dB too loud...

With the manLOUD process the Smyths already implemented a very cool and handy feature which I think many people don't yet understand completely.

No I did not make a manLOUD HPEQ for the Surrey room, but now that I know about it I will experiment. I’m not 100% sure that I understand the procedure that I need to follow, but maybe with some trial and error I’ll get it figured out.

 

[audiohobbit]

Do you use the "tri-vol" option ? Appendix 25

Because you can store 3 volume levels by preset. You recall those levels with the rocker buttons on the remote.

Good question!
After the FW update for 1.75 several options were enabled/set to different settings and the Tri Vol option was also activated. This is why you should do the Restore factory setup (accidentally I did the full factory reset. A bit confusing that there are 2 similar named options).

Another addition to the wish-list would the possibility to save listening rooms and presets to SD card, if possible. Especially setting up a listening room and saving it could take a long time.
With the A8 you could save presets to SD card (The A8 has no separate listening rooms).

 

[audiohobbit]

No I did not make a manLOUD HPEQ for the Surrey room, but now that I know about it I will experiment. I’m not 100% sure that I understand the procedure that I need to follow, but maybe with some trial and error I’ll get it figured out.

I will write a tutorial for that, but I'm not sure if I will find time for this this week.
The video is already a good starting point.
Do you at least understand the theory behind this, which I tried to explain in my previous post?
I'd recommend using a flat HPEQ as a starting point. So generate a flat HPEQ and store it to the internal memory.
Then go again to the HPEQ app, and follow the video in selecting equal loudness 80 as target curve ("Curve"). Man EQ Start should then be flatEQ (as opposed to the video they propose to start with an auto HPEQ but I'd recommend a flat HPEQ).
In Man EQ HPEQ you selsct the flat HPEQ you just created and stored.
In Man EQ PRIR you select the (factory or any other) PRIR you want to personalise.
You swithch on Man EQ speaker and select one of the 3 options I mentioned. I'd start with L+R.
In HP (run) you select manLOUD and press enter. Then you heare the "base band" 0-500 Hz. You have to level this to about 80 dB. You just have to guess this.
The rest is explained very good in the video. You have to got through the sub bands back and forth and also cycle fast through them then you'll notice loudness differences. Most important are tha bands up to 5-6 kHz.
When I set all the levels right I go through all 32 bands again and store a picture to the SD card, so I have 32 pictures at the end per manLOUD HPEQ (it's written elsewhere in this thread how to store pictures and how to activate this first in the Factory Test menu otherwise it won't store pictures). Most important is at the end to press the HP key! Only then your manLOUD HPEQ will be saved. It will not be a separate HPEQ file but an added option to select in the (flat or auto) HPEQ file you created earlier!
So if you want to make a new manLOUD HPEQ you have to create a new flat HPEQ first, because only one manLOUD HPEQ can be stored in one file together with a flat or auto HPEQ.
Problem is you can't give those files indiviual names. They all have the name of the user and headphone you entered in the System menu.
So you only can distinguish them by date and time created. The date and time are those when the basic flat (or auto) HPEQs are generated, not when the manLOUD HPEQ was created as an addition to one of these files.

And when you store such an HPEQ to SD card, the filename gets the date and time when the file was written to SD card, not when the HPEQ was created. The creation dates and times are only written in the file header. You can open the HPEQ file with a text editor, the first few lines are text with user and HP name and date and time of creation. The rest is binary and looks like garbage in a text editor.
For each of those manLOUD HPEQs (or any other HPEQs) I make a folder on my PC with date and time of creation, what it contains (manLOUD_Surrey for example) and so on. I copy the HPEQ file to this folder and the 32 pictures of the subbands I stored for this manLOUD HPEQ. So you can replicate this later. Otherwise it's a black box and you have no idea what you set.

 

[SobiC]

I'm not sure that a simple recording of the audio sweeps would work.

It works flawlessly. The timing within the recording has to be unaltered and the first impulse is crucial. I've done several measurements, the only time it didn't work was when I pressed the SPK button slightly too late.

 

[sander99]

Yes it definitely does. The manuals says:

The HPB→A mode, while enabled, also correctly switches the headtracking input signal from User A to User B.

"while enabled" could mean as long as this option is switched on in the system menu, not only when actually rerouting B to A via arrow key in speaker map mode.
I tried this yesterday and it seems that it is so. So you have to switch that off in the system menu ( It's now in Misc Settings with the new FW). It's not meant ot be on all the time I think.
So, no bug.

@jgazal ("cc" @Erik Garci): so the user A and B headtracking can be linked, that's one step in the right direction for my dual user cross talk reduction/cancellation idea. (Now we still need individual speaker level control per user. Or try to create dummy silent speakers or find some other trick.)

 

[audiohobbit]

It works flawlessly. The timing within the recording has to be unaltered and the first impulse is crucial. I've done several measurements, the only time it didn't work was when I pressed the SPK button slightly too late.

But you should end up with a lot of delay in those PRIRs. In synchronous mode the Realiser starts to play the sweep and expects the recorded sweep ideally to start at the same time. If it starts later, then that would be delay (ie speaker distance). I can't imagine that you can press a button within less than say 100 ms.
So you should get 100 ms or even much more delay in those PRIRs that adds up to the processing delay the Realiser already have.

By the way: Yesterday I played an Atmo Bluray and the 90 ms delay I set with DTS->PCM worked fine. So I still can not report a difference between Dolby processing and PCM (non-) processing.

My projector should have a delay of video of about 140 ms. That has been measured in one test. This would mean my Realiser would have a processing delay of about 50 ms, since I have to set 90 ms extra delay (in the player or the Realiser) for the audio.

 

[Dixter]

It works flawlessly. The timing within the recording has to be unaltered and the first impulse is crucial. I've done several measurements, the only time it didn't work was when I pressed the SPK button slightly too late.

Now we just need to get the 9.1.6 and the rest of the sweeps...

 

[Got the Shakes]

I will write a tutorial for that, but I'm not sure if I will find time for this this week.
The video is already a good starting point.
Do you at least understand the theory behind this, which I tried to explain in my previous post?
I'd recommend using a flat HPEQ as a starting point. So generate a flat HPEQ and store it to the internal memory.
Then go again to the HPEQ app, and follow the video in selecting equal loudness 80 as target curve ("Curve"). Man EQ Start should then be flatEQ (as opposed to the video they propose to start with an auto HPEQ but I'd recommend a flat HPEQ).
In Man EQ HPEQ you selsct the flat HPEQ you just created and stored.
In Man EQ PRIR you select the (factory or any other) PRIR you want to personalise.
You swithch on Man EQ speaker and select one of the 3 options I mentioned. I'd start with L+R.
In HP (run) you select manLOUD and press enter. Then you heare the "base band" 0-500 Hz. You have to level this to about 80 dB. You just have to guess this.
The rest is explained very good in the video. You have to got through the sub bands back and forth and also cycle fast through them then you'll notice loudness differences. Most important are tha bands up to 5-6 kHz.
When I set all the levels right I go through all 32 bands again and store a picture to the SD card, so I have 32 pictures at the end per manLOUD HPEQ (it's written elsewhere in this thread how to store pictures and how to activate this first in the Factory Test menu otherwise it won't store pictures). Most important is at the end to press the HP key! Only then your manLOUD HPEQ will be saved. It will not be a separate HPEQ file but an added option to select in the (flat or auto) HPEQ file you created earlier!
So if you want to make a new manLOUD HPEQ you have to create a new flat HPEQ first, because only one manLOUD HPEQ can be stored in one file together with a flat or auto HPEQ.
Problem is you can't give those files indiviual names. They all have the name of the user and headphone you entered in the System menu.
So you only can distinguish them by date and time created. The date and time are those when the basic flat (or auto) HPEQs are generated, not when the manLOUD HPEQ was created as an addition to one of these files.

And when you store such an HPEQ to SD card, the filename gets the date and time when the file was written to SD card, not when the HPEQ was created. The creation dates and times are only written in the file header. You can open the HPEQ file with a text editor, the first few lines are text with user and HP name and date and time of creation. The rest is binary and looks like garbage in a text editor.
For each of those manLOUD HPEQs (or any other HPEQs) I make a folder on my PC with date and time of creation, what it contains (manLOUD_Surrey for example) and so on. I copy the HPEQ file to this folder and the 32 pictures of the subbands I stored for this manLOUD HPEQ. So you can replicate this later. Otherwise it's a black box and you have no idea what you set.

I do understand the concept behind what you’re saying. Having only done the autoEQ process before I’ll obviously have to play around with this and see if I can get it sounding right without a step by step guide. The good thing is it’s just a bunch of 1s and 0s and if I screw it up I can start over pretty easily.