Long awaited Smyth SVS Realiser NOW AVAILABLE FOR PURCHASE

[darinf]

Quote:

 
Ok; I notice that the accuracy of the emulation is very dependent on the mic fittings, and despite me having done it almost a dozen times myself now I still don't have it down to a science. I'm going to ask Lorr to give me some pointers on that in person, hence the LA locations (where he's able to go help).
 
Once I get good at putting the mics I'll consider going to other cities.

I placement of the mics is VERY important. I have done a LOT of measurements for me and others. I have also done experiments with different foam tips, different length tips, super deep insertion of the mics, etc. I did these measurements all on the same system and compared the results. I also take photos of the mic placement for each measurement I do so I can see if there is any specific mic placement issues and how it affects the accuracy of the resulting measurement when I am comparing the results.
 
That's why people have made the trip to San Diego to have me do their measurements. I know how to get the most accurate measurements, in some great sounding rooms with very high end equipment.
 
AIX is a great place to start, but personally, I find recording studios tend to be very "damped" and "dead" sounding. I prefer a more "lively" sounding room, especially for listening to music. But others certainly like the sound of AIX.

 

[YCH]

Quote:

I placement of the mics is VERY important. I have done a LOT of measurements for me and others. I have also done experiments with different foam tips, different length tips, super deep insertion of the mics, etc. I did these measurements all on the same system and compared the results. I also take photos of the mic placement for each measurement I do so I can see if there is any specific mic placement issues and how it affects the accuracy of the resulting measurement when I am comparing the results.
 
That's why people have made the trip to San Diego to have me do their measurements. I know how to get the most accurate measurements, in some great sounding rooms with very high end equipment.
 
AIX is a great place to start, but personally, I find recording studios tend to be very "damped" and "dead" sounding. I prefer a more "lively" sounding room, especially for listening to music. But others certainly like the sound of AIX.

 
What are the dominant factors you have found to produce the most consistently + accurate emulation results?
 
I guess symmetry would be one. I think I get better results with deeper mic locations too. Would like to hear your take on this.

 

[darinf]

Quote:

 
What are the dominant factors you have found to produce the most consistently + accurate emulation results?
 
I guess symmetry would be one. I think I get better results with deeper mic locations too. Would like to hear your take on this.

Symmetry is tough because many people do not have symmetric ears. But, yes, the closer the better when you are trying to match the left and right mic placement.
 
We tried cutting down some custom foam tips and inserting the mics really deep into the ear canal. Probably not the safest thing to do, but the mics were far enough in that it was hard to see that the mics were perpendicular to the ear canal. The measurements from the very deep insertion was not good at all. In fact it was terrible! That result was counterintuitive, but I suspect that the Realiser recording was never designed for such deep insertion of the mics.
 
As far as the mic placement in general, it's nothing you wouldn't already guess. In my opinion, It's best to have the mics flush with the beginning of the ear canal. Not sticking out, but not too far in either. Also, the mics have to be perpendicular to the ear canal. Any angling towards the front or back seems to affect the measurement. It's really kind of a judgement call.When I look at the mics inserted, I can just tell if it's in the right place.
 
Even with all that, sometimes one measurement just sounds better than another. There are so many factors to getting that last few percent of accuracy.
 
I don't know if Lorr would agree about how much any of this matters. But it's also in their best interest to tell people how easy it is to get a good measurement with the Realiser. You don't want potential customers thinking that it's too hard to get a good measurement or that they won't be able to do it.
 
I don't think it's hard to get a good measurement with the Realiser. Yes, you have to know what you're doing and have  to read the manual. But I think most people could do a good measurement. But like anything having to do with high end audio, it does take some work and skill to be able to squeeze all the performance you can out of a given piece of gear.
 
For example, with speakers, you can buy some really good speakers and set them up in your room and they will probably sound pretty good. But, if you educated yourself on speaker placement and trained yourself how to hear if there are problems, etc. then you could usually make the speakers sound a lot better. You could use a similar analogy to buying a video projector or large screen monitor. Most high end owners assume they will spend money having a certified technician come in and tweak their gear. It doesn't mean the monitor or projector will look bad out of the box, but it can certainly look a lot better in the right hands.

 

[Erik Garci]

Quote:

We tried cutting down some custom foam tips and inserting the mics really deep into the ear canal. Probably not the safest thing to do, but the mics were far enough in that it was hard to see that the mics were perpendicular to the ear canal. The measurements from the very deep insertion was not good at all. In fact it was terrible! That result was counterintuitive, but I suspect that the Realiser recording was never designed for such deep insertion of the mics.

Were the mics inserted very deep when measuring the headphones (HPEQ)?  Or only when measuring the room/speakers (PRIR)?  Maybe they should be inserted the same way for both the HPEQ and the PRIR.

 

[darinf]

Quote:

Were the mics inserted very deep when measuring the headphones (HPEQ)?  Or only when measuring the room/speakers (PRIR)?  Maybe they should be inserted the same way for both the HPEQ and the PRIR.

I found that it's always best to do an HPEQ every time along with the PRIR. This way the mic placement is the same for both. So, yes, we did the HPEQ right after the PRIR without moving the mics.
 
Even when not doing a deep insertion test, I always do a HPEQ right after the PRIR. I then keep the PRIR and HPEQ as a "matched set". Good thing I only have one pair of headphones, otherwise I would feel compelled to do an HPEQ on all my headphones for each measurement.

 

[hekeli]

Darin, have you still done any manual EQ routines? Or do you still think that Realiser has somehow better ears than your own?

 

 

[darinf]

Quote:

Darin, have you still done any manual EQ routines? Or do you still think that Realiser has somehow better ears than your own?

 

Interesting way to phrase the question.
 
First, yes, I have done the manual EQ before. However, most places I do measurements barely give me the time to come in a do a measurement, let alone do a manual EQ.
 
I still don't trust that my ears are consistently accurate enough to do the EQ properly.
 
Do I think the Realiser has better ears than I do? Yes!
 
I think of something like monitor calibration. Do you think a spectrophotometer and a piece of software has better eyes that I do? Well, for measuring brightness levels and color, yes. If it wasn't better, then we would all just use our eyes to color calibrate our monitors. (Some people do!) But would you trust a certified calibration technician if they came in to calibrate your monitor and just used their eyes to tweak your monitor or projector? You would think they are crazy.
 
Are microphones and measurement software better than my ears? Yes! I cannot identify say a -2dB dip at 1520kHz, but a microphone and software can. Most people trust test equipment more than their own senses for some things.
 
Yes, ultimately it is your ears and your brain that matters, but our ears and brains are easily fooled and rarely very consistent.
 
Another example, I color correct photos and video for a living. I can adjust a photo looking exactly how I want it and how I remember the scene when I shot the photo. But then I can come back 2 hours later and look at the same photo and wonder what was I thinking making the photo so red or so blue, etc. I will have my blue days, my red days, etc. Maybe I am just not very good at it!
 
All I am saying is that I don't trust my ears. I should do an experiment and do a manual EQ one day and then do another one a week later and see if they come out the same.

 

[hekeli]

Yeah you have some points like it takes a good while. But if you believe that even HPEQ should be done at the same time as PRIR, it's pretty simple to conclude you should really do the manual EQ too at the same time. After all the A/B is pretty much instant and you _will_ hear the differences easily, it's the only way to guarantee 100% if that's what you are looking for.
 
Has anyone managed to do a basic PRIR and A/B it right there sounding identical without a shadow of a doubt? From what I experimented, manual EQ fixed nicely some difference in highs etc - small changes but easily audible. Of course this might or might not matter to someone. If you are sparse on time, it might be better to just rapid fire as many PRIR's as you can and listen to the best sounding one to you.

 

[Blackmumba]

Unfortunately I had to sell my Realiser and Stax SR-007, I was getting major ear buzzing from using headphones in my left ear.

 

[jk6661]

Sorry if this has been addressed, but...
 
Is anyone bothered by the fact that the Realiser processes at 24/48? That means you can use the Realiser digital out to feed the best DAC on earth, but you're still only hearing 24/48. I tried convincing someone at Smyth to upgrade the Realiser DAC to 24/96 at least for the two front channels, but it seemed to fall on deaf ears. Seems chintzy for a $3000 piece of equipment. That said, I still enjoy mine quite a bit.

 

[darinf]

  Sorry if this has been addressed, but...
 
Is anyone bothered by the fact that the Realiser processes at 24/48? That means you can use the Realiser digital out to feed the best DAC on earth, but you're still only hearing 24/48. I tried convincing someone at Smyth to upgrade the Realiser DAC to 24/96 at least for the two front channels, but it seemed to fall on deaf ears. Seems chintzy for a $3000 piece of equipment. That said, I still enjoy mine quite a bit.

Well, now you're getting into the ongoing debate about high resolution music. I won't even go there, but let's assume that more is better..
 
As we know, the Realiser is a unique device and increasing the data processing rate is non-trivial. It's not a matter of just upgrading a DAC chip that supports 24/192K.
 
First, the processor is processing up to 8 channels of audio. So it would have to be upgraded to have enough processing power to process say 8 channels of 24/192K audio in-real time with very low latency. That is a LOT of processing. Even if the high data rate could only be used on 2 channels, that's still a decent amount of processing. I have no idea, but I am guessing they would have to upgrade the processors in the Realiser and write new code to process at that data rate.
 
Secondly, feeding 24/192K data into the Realiser for processing would require a 24/192K PRIR measurement. So now you're talking about supporting 24/192K on the HDMI input AND having a 24/192K A to D converter. Then you also have to update all the code to process the high bit rate data to create the "high res" PRIR.
 
So, unless you do all of these things, the only other way to do it would be to just upsample everything at the output. But you can do that now if you have a DAC that upsamples.
 
Let me qualify that I have no inside information about the Realiser. I am just theorizing at this point. But it makes sense to me.
 
Who knows? The Realiser has been available for a while now. Maybe they have something in the works? My point is it's non-trivial to say the least. And that assumes that they believe higher resolution will yield better sound.

 

[blueangel2323]

They should create a cheaper "Lite" version that has a few fixed profiles of some of the best speaker setups and listening rooms in the world, without the ability to create custom profiles of your own listening rooms. Not everyone has a high end, properly calibrated speaker system, set up perfectly, in an acoustically ideal room, that they can use to measure. I don't, and if I did, I would be listening to that instead. The lite version could also do away with the head-tracking, which IMO is a gimmick. I just want a realistic, speaker-like soundstage and presentation in a pair of headphones without spending a few grand.

 

[Erik Garci]

  First, the processor is processing up to 8 channels of audio.

 
In dual user mode, it is effectively processing up to 16 channels of audio (8 per user). So maybe there is already enough processing power for higher resolutions.
 

  Secondly, feeding 24/192K data into the Realiser for processing would require a 24/192K PRIR measurement.

 
Maybe the existing 24/48k PRIR could be upsampled to 24/96k or 24/192k, instead of measuring a new PRIR.
 
Like you said... who knows?

 

[jk6661]

  Well, now you're getting into the ongoing debate about high resolution music. I won't even go there, but let's assume that more is better..
 
As we know, the Realiser is a unique device and increasing the data processing rate is non-trivial. It's not a matter of just upgrading a DAC chip that supports 24/192K.
 
First, the processor is processing up to 8 channels of audio. So it would have to be upgraded to have enough processing power to process say 8 channels of 24/192K audio in-real time with very low latency. That is a LOT of processing. Even if the high data rate could only be used on 2 channels, that's still a decent amount of processing. I have no idea, but I am guessing they would have to upgrade the processors in the Realiser and write new code to process at that data rate.
 
Secondly, feeding 24/192K data into the Realiser for processing would require a 24/192K PRIR measurement. So now you're talking about supporting 24/192K on the HDMI input AND having a 24/192K A to D converter. Then you also have to update all the code to process the high bit rate data to create the "high res" PRIR.
 
So, unless you do all of these things, the only other way to do it would be to just upsample everything at the output. But you can do that now if you have a DAC that upsamples.
 
Let me qualify that I have no inside information about the Realiser. I am just theorizing at this point. But it makes sense to me.
 
Who knows? The Realiser has been available for a while now. Maybe they have something in the works? My point is it's non-trivial to say the least. And that assumes that they believe higher resolution will yield better sound.

 
I was only talking about 24/96, at least for starters. Also, I mentioned processing 24/96 only on the front two channels, which arguably would benefit the most, depending on your view of hi-res music. (I agree, let's not go there.) Considering you can get 24/96 DACs these days for a couple of hundred bucks, I really don't think I'm asking for miracles at a $3000 price point. 

 

[arnaud]

I guess I could see interest for an update which handles 24/96 all the way through in case of stereo input. It may be hard to justify for smyth though as their core audience are people mixing/mastering else consumer of multi channel movie soundtrack. It seems DTS-HD MA can support 5.1 at 192kHz or 7.1 at 96kHz for BR disc so, while I haven't check what proportions of BR material are offered with high res multi channel soundtracks, smyth would probably need to handle this if they're to address their core market and not just a niche within the niche that is the headphone buff.

Besides, I am not sure how many of the pro-audio customers of smyth do care that their multi channel soundtrack is decimated (litteraly as you go from 192 or 96 down to 48, that is a decimation rather than resampling) for use with the realiser. As I imagine the realiser is used by these people as had-oc monitoring system when they don't have access to their studio, it might be just fine in its current state to serve that purpose...

Now to the main point, assuming this is implemented based on a resampled PRIR, do you think there's still benefit to gain from? E.g., are you benefiting from using an upsampled prir/hpeq rather than decimating the input stream as currently done?

If the request is to also upgrade the prir recording process to 96khz (or even 192kHz), it does bring interesting practical questions such as:

1. Is the speaker system I am measuring going to output anything above the audio range?
=> I guess with modern amplfiers and speakers with diamond and other ceramic tweeters, it's possible.

2. Is the in-ear microphone I am using going to capture anything above the audio range?
=> Just the foam used to sit the mic in the ear canal provides pretty much 100% sound absorption by the time you reach 20kHz.

3. Will my PRIR be representative / usable for frequencies above 15kHz?
=> Half an acoustic wavelength is about 8mm at 20kHz.
=> Tilt your head / move your torso by a 1/6 to 1/3 of an inch during the prir recording and all bets are off as you'll be trying to track an constantly varying response as you can't hold your position that steady.

4. Assuming ultrasonic content was effectively recorded in the prir, will my brain use that part for the localization?
=> I don't know enough detail about our hearing system to answer the question, I suppose it does just the same as in the upper part of the audible range (mainly inter-aural level difference)