That makes sense its not like ART had a bad PET in the graph I posted earlier it is essentially flat from 45hz and above. I'm guessing that is the additional speakers being leveraged. Also my spectrogram of my manual system was only from one mic position; it would be better to see spectrogram of a basic manual calibration (acoustic timing reference/subwoofer time aligned) vs Dirac ART at different mic positions. Obviously Dirac ART is way more advanced than what I am doing so I am guessing it would probably perform better off axis than my system. So even from a time spent calibrating sense; it is a good trade off to just go with Dirac ART.
-
Welcome to ASR. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!
miniDSP Tide16 - Holy Grail with 16 Channel Atmos/DTS:X, high SINAD
welwynnick
Addicted to Fun and Learning
- Joined
- Dec 26, 2023
- Messages
- 944
- Likes
- 1,188
I think that's intentional - it doesn't even try to.
Firstly, infrasonic sound waves have long wavelengths, so for domestic rooms, they're hardly travelling waves any more.
The sound pressure around the room will be close to being in phase, so there's little room correction to be done.
Secondly, the filters would have to be very long, so the DSP would invoke an unacceptably long audio time delay.
Having said that, I'm sure I read (maybe incorrectly) that ART doesn't just stop correcting below 20Hz, it rolls off altogether, which doesn't sound right to me.
From what I've seen and heard, compared with the rest of the frequency range, ART seems to rein in the low frequency reverberation time unnaturally tightly.
The "waterfall" from 20 to 150 Hz looks like a cliff instead.
It's hard to look at graphs all the time. Better to enjoy ART in person!
One practical advantage of the Tide 16 is that you can maintain multiple scenes, for example a manual, MLP-optimized calibration and a Dirac ART scene, and switch between them instantly. I'm thinking there has to be some kind of tradeoff if you have to use other speakers to cancel reflections of another speaker so just use Dirac ART strategically. Multiple speakers and subwoofers will have different wavelengths and phase rotations and phase consistency at any given frequency is going to be dependent not just on calibration method but timing and location. Optimizing for one seat will change the pattern of cancellations and summation elsewhere regardless of the calibration method. Have friends over or family movie night use.. Dirac ART. Critical listening use your manual calibration for one seat.
From 20hz-150hz is ART’s range.
Also this to me means that my surround speakers need to be good quality, ideally just as good as the main speakers. Which means I need to upgrade my surrounds sooner than later.
Keith_W
Master Contributor
There are many in that Dirac ART thread. But here is one from my collection. I didn't take the measurement, I downloaded it from ASR. It's a bit large, so I can't upload it. I have hosted it on my Google Drive here. Grab it quick, I will delete it in a couple of days.
You will see that it is a collection of individual speaker and sub measurements of DLBC vs. ART, all with a timing reference. First, the good:
This is a comparison of the vector sum of all DLBC measurements (dark yellow) vs. ART (blue). I wouldn't call this particular example a great result, I have certainly seen better measurements in that ART thread. But let's roll with it for now.
This is the spectro, normalized to peak at each frequency, for DLBC (left) vs. ART (right). Note the vertical and horizontal scales are the same. IMO this result is genuinely impressive. Now, the bad.
Step response of the subwoofer and the vector sum for DLBC. Note how delayed the sub is, and in this user's case, there is a giant reflection occurring at about 60ms. I have no idea what DLBC was "thinking" when it did this.
If we look at the individual responses of each speaker with DLBC, we can see that none of them have any "pre-ring" before the main impulse at 0ms.
Comparison of the step response of the vector sums of ART (yellow) vs. DLBC (blue). We can see that there is a substantial portion of the step that occurs before 0ms. Given how loud it is, I would think it should be audible. But then it's low frequency and starts at -20ms. I don't know what the pre-masking threshold is at low freqs. Do you have a link by any chance?
More interestingly, if we look at the step response of every speaker in ART, we can see that it works by making the "support" speakers play low frequencies before the upper frequencies!
DLBC channels are all over the place but ART channels basically look like a single step response. ART is paying very close attention to timing/phase. ART is a lot hotter in the bass, I'm guessing it is a multiple subwoofer system and they are getting very good summation but they might be boosting ... You still see the lingering bass energy in the spectrogram but it slopes down at 40hz (I think this is the support speaker cancellation effect). Since the subs dominate the vector sum I think that means they use a steep HPF on the mains. Basically it's saying crossover frequencies should be sub dominant. That is a safe decision because you would want the driver better suited at those frequencies to dominate. ART is impressive. The lingering bass energy I think you could say is perceptual preference though and may not be desired by all. I think the decision making they made is what is safe for any system that might be using this. If I had better front speakers with strong low frequency response I personally would use a gentle HPF. Dirac ART is like a one size fits all type design.
- Joined
- Jul 12, 2024
- Messages
- 2,209
- Likes
- 3,277
Right. I run a 2ch headphone off my HTx that I use for 5.1.
- Joined
- Jun 5, 2016
- Messages
- 3,327
- Likes
- 5,637
We all have our own hangups, but that’s not the reality. While timber-matching (spelling intentional to reflect reality as opposed to marketing pablum) may scratch an aesthetic itch, the role is different. Even when you have things like Mahler 8 brass on the balcony, human hearing works differently from the sides and backs than from the front. IME, one wants wider dispersion in the mids and lower treble, the top octave is irrelevant, and an array that falls off at less than 6dB per doubling of distance is beneficial.
This evening I’ve been enjoying a number of immersive recordings* — currently Abbey Road over an immersive system that includes identical constant directivity 2-way LCR speakers of medium pattern width, sides and rears that have 16 2” wideband drivers in a shaded CBT array, and front/side-rearish heights that are mini-mes of the LCR (largely for timber matching, lol, though they also have more headroom than pretty much anything else their size), along with 4 high power subs in the 4 corners of the front wall, all stitched together with ART. The immersive bubble is as good as or better than anything I’ve heard that wasn’t in the Philharmoniker, Musikverein, Woodruff Hall, or Chicago Orchestra Hall. Even though if one set up the sides and rears in front they would be markedly inferior performers than the current LCR.
Now, I do think more capable surrounds and heights than one typically uses is probably smart with ART, given the inherently higher demands placed on the ancillary speakers. That’s one reason why I went with bigger side speakers in this room than in our previous home, and potent mini monitors with in room extension below the room’s first mode and high output with low compression for the heights.
*I will note this room is not at all soundproofed, and next to a central square-shaped staircase. So levels are low. Yet, ART and well-judged loudness compensation make for extraordinary resolution even at rather low levels.
Judging by that ART step response post above, it looks like the built in crossover behavior is a fairly steep, sub dominant. From what everyone else is saying it treats 20 to 150 Hz as one big system and, based on the measurements, decides at each frequency how much energy should come from the subs, the LCR, and the surrounds or heights. You can move around the room and still be inside the same immersive bubble because ART is managing timing and low frequencies for all of your listening positions.
If you ever want a pure single seat preset instead of ART, DSP gives you freedom to experiment. Example: LCR with Bessel 12 dB per octave high pass around 60 Hz so they keep some low frequency, the CBT surrounds and heights high passed with LR 24 db per octave around 100 Hz so they are not doing heavy bass, and the four subs low passed with same slope as LCR. It's fun to experiment with different crossovers in my experience depending on what type of music you are listening to. You can squeeze more out of one listening position, but would have to give up that immersive bubble ART is giving you.
3ll3d00d
Senior Member
- Joined
- Aug 31, 2019
- Messages
- 416
- Likes
- 303
fwiw the link is a target curves zip so I guess it's the wrong link
https://www.bodziosoftware.com.au/Pre_Post_Ringing_IR_And_Pulses.pdf is the only thing I recall reading on that topic which also references ~20ms
Sokel
Grand Contributor
- Joined
- Sep 8, 2021
- Messages
- 11,128
- Likes
- 12,074
I may be misreading the reply but there is a wink at the end so maybe?R.I.P.
(about FIR of course)
Sokel
Grand Contributor
- Joined
- Sep 8, 2021
- Messages
- 11,128
- Likes
- 12,074
It says clearly, "no custom FIR at first built" .
If they built some other version or include it some time afterwards I promise I'll put a nice shovel at work
Ultrasonic
Addicted to Fun and Learning
I considered a UMIK-2 for stereo Dirac LIve but decided against it due to the inferior angular response. If I wanted to make distortion measurements I'd want one over a UMIK-1 but for fundamental frequency measurements I didn't see a plus from the lower noise floor, and I don't believe Dirac will be trying to do anything with harmonic content.
Is there maybe some other consideration where you've seen a benefit of a UMIK-2 though?
Sokel
Grand Contributor
- Joined
- Sep 8, 2021
- Messages
- 11,128
- Likes
- 12,074
The main one is clock-drift.
That's what messing with timing and it's needed for FSAF for example (bare minimum, REW suggests higher-end mics for this) .
If you read me fixed to this test, is just because blows up the "what about with music? " subjective argument as one can use just about anything as test signal and do both electric and acoustic measurements.
Agreed, for someone who don't need more than 3 HDMI inputs, 1 HDMI output, HDMI 2.1/8K, phono input, but do need USB UAC2 input, with all Dirac Live licenses included; and like the feeling of having an AVP that will likely measure as good or better than many dedicated two channel stereo integrated, or separate preamp/power amps, this $3,500 devices is an excellent option. In fact, it seems like the only option available at the moment.