1. This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
    By continuing to use this site, you are consenting to our use of cookies.

    Dismiss Notice

Objectivists board room

Discussion in 'Sound Science' started by joe bloggs, May 28, 2015.
254 255 256 257 258 259 260 261 262 263
265 266 267 268 269 270 271 272 273 274
  1. Argyris Contributor
    DSP is inevitable, and I welcome its eventual arrival. Contrary to what the people blowing multiple thousands on each new flagship might say to justify their purchase, there are no massive gains to be had with passive drivers anymore. The same basic issues--spikes in the lower treble, lack of absolute bass extension in open dynamic designs, raggedness in treble response and a wide peak around 1 kHz in planar and electrostatic designs, the incorrect placement of the presence hump at ~2.5 kHz instead of 3.5 kHz in IEMs, the strong 5 kHz spike also in IEMs, and on and on--these same issues keep coming up over and over and over again in every new product.

    At this point, we have to accept that, without active correction, passive designs will always have at least some of these issues. Why spend exorbitant amounts of money and effort playing whack-a-mole trying to engineer a solution to one of these things, maybe not even completely alleviating the issue or else introducing another issue while solving the first, when you can take a driver with workable frequency response and good time domain characteristics (i.e. no ringing) and actively correct the FR. What's more, you can even measure the system (i.e. the headphone while it's on the individual listener's ear) and, in real time, generate precise corrections to hit any target curve. Want a different sound, or a closer approximation of neutral as further research refines the target curve? Just recalibrate for a different curve.

    The beauty of all this is that, since the mainstream industry is already moving toward wireless headphones, you're already building an amp and some IC into the cups, anyway, so how much extra effort would it be to fit in a DSP solution on the PCB and cobble together an app to program it? The effort, of course, would be in designing the calibration system, choosing a curve, etc. But my point is it would fit in the cups alongside the stuff you're already putting in them to power the headphone.
    Last edited: Sep 26, 2017
  2. bigshot
    I don't even think that a lot of what they call time error or ringing is audible. Just fix the response and you're almost all the way there. The rest can be fixed with good recordings.
  3. briskly
    • Limited bass extension is the simple result of intentional leakage integrated into the headphone.
    • Treble response issues in headphones are so very much like bass response in rooms. The dimensions are on the order of the wavelength. Subject to multipath interaction with pinna, blah blah
    • I thjnk you follow innerfidelity too closely. Mean adult canal resonance at around 2.7kHz. His use of ID tends to overemphasize 1 kHz region in compensated result.
    • IEMs must assume transfer function of outer ear, specifically the pinna and open canal, and recreate it. New closed canal resonance is a function of depth from eardum.
    • Time and frequency domain issues are one and the same.
    Last edited: Sep 27, 2017
  4. headwhacker
    What exactly do you mean by leakage in the headphone leading to limited bass extension? The iSines are open back and doesn't require a tight seal from the tip, but the bass is well extended.
  5. Niouke
    I have XBA-3 IEM's with the atrocious 5Khz bump, but I thought it was specific to the driver choice and tuning from sony?
  6. Argyris Contributor
    1. In an open headphone? Wouldn't the open back make any intentional leakage moot? I know that the trend lately is for closed designs to have intentional leakage built in so as to decrease reliance on a tight seal, but I was of the understanding that dynamic drivers have poor bass extension in open enclosures because they have trouble producing output below their resonant frequency.

    2. Above about 10 kHz, this is certainly true for every headphone, and it's also outside the guaranteed area for measurements so it's probably safe to ignore what happens here unless the average output is markedly deviant from the norm. Below that, though, there are headphones that measure reasonably smoothly and ones that don't. Maybe the average trend line is the most important thing, but there's certainly better and worse performance here.

    3. The source of measurements doesn't matter, as long as the measurements are internally consistent. Tyll's raw measurements universally show the peak at ~3.5 kHz in headphones, and at ~2.5 kHz in IEMs. For that matter, all the raw measurements I've seen elsewhere also have the peak at around 3.5 kHz in headphones, so if it's actually supposed to be 2.7 kHz, then all these measuring rigs have the same error. And even if they do, this is irrelevant because the IEM measurements are still offset by about 1 kHz from the headphone ones, and it's this offset--not the actual numbers--that's important. Significantly, Tyll's dummy head stuck in a room with speakers measured the peak around 3.5 kHz, implying that where it places this peak for headphones corresponds to what a listener actually hears with speakers or with live music. The IEMs are the odd one out here.

    3.5. The 1 kHz bump in electrostats and planars is in comparison to other headphones, which generally don't have this feature. The compensation shouldn't matter, as the difference exists regardless of the compensation used. One may of course argue that this level of energy around 1 kHz is actually correct, and that headphones that don't have it are less accurate in this area. Tyll's new compensation curve incorporates this very assumption, in fact. However, that of course is debatable and it nonetheless doesn't obviate the existence of the 1 kHz lift as a distinct feature shared by electrostats and planars and rarely seen elsewhere.

    4. Indeed, this is the difficulty in getting an IEM to sound correct. Unfortunately, because of the physics involved with shortening and stopping up the ear canal, they struggle. Additionally, an individual listener might prefer a deeper or shallower insertion depth or different tips from the ones the IEM was designed and tuned for, and the depth and fit might vary significantly each time the IEM is used. These factors will create variations in performance from user to user and from session to session. It's a difficult set of problems to overcome, and it's kind of amazing that any IEM manages to sound good.

    5. I'll defer to the consensus of the thread here, since this is a topic I could stand to learn more about. I'll submit some points for discussion, though. In my understanding, treble spikes generally produce ringing, but at least in the case of the DT880, this doesn't seem to be the case. It has significant spikes at ~6 kHz and ~8.5 kHz, but I've seen CSD charts (elsewhere, in places unmentionable) that show it having a smooth, even decay across its entire frequency range. Additionally, its square wave measurements have only a small section of clear ringing at the beginning, which resolves to a flat line relatively quickly. Its measurements here mimic those of the HD 6x0 family, which does not have any significant treble spikes. Is there perhaps something going on here that has nothing to do with the treble, that's similar between the two headphones and which produces such similar square waves? Why don't the DT880's spikes ring the way they do on many other headphones? And, probably the most important question, how much of this can we actually hear?
    Last edited: Sep 28, 2017
  7. Argyris Contributor
    If you look at the raw measurements for a lot of IEMs (virtually all that I've seen, in fact), they all have the ~5 kHz spike in some capacity. I believe it's an ear canal resonance, since it's right around double the frequency of the presence bump lower down, implying a relationship. Whatever causes this spike, some manufacturers try to compensate for it by tuning the entire area around it down, multiple driver models seemingly taking advantage of a crossover zone to do this. Others just leave it there, apparently figuring it's best left alone. Still others, like Etymotic, seem to have found a way to tune it downward in frequency and integrate it into the presence bump, perhaps in the latter case because of their use of deep insertion triple flange tips. In my experience, generic triple flange tips on my (now unfortunately lost) EX1/Titan 1 had the same effect--the telltale hyped sibilance and edginess from the spike at 5 kHz disappeared, and the tonality definitely reflected more energy in the 3.5 kHz to 4.5 kHz range. I'd love to see comparative measurements with different tip types to confirm what's happening here.
    Last edited: Sep 28, 2017
  8. headdict
    A tight seal is absolutely needed despite the open design. A significant amount of sub bass is lost when I use some of the smaller tips. Only the largest size gives me a good seal. Try for yourself...
  9. Zapp_Fan
    As someone working in the mainstream industry, I can confirm this. Many if not most decent BT chipsets already include DSP with EQ and some other common effects. Nothing too interesting like convolution or even cross-feed yet, but EQ is already here and fancier stuff will follow once EQ and surround effects and suchlike become common built-in features.

    Of course this means you're stuck with BT audio which is not as high-bandwidth as I'd like, but it's a path towards a lot more flexibility and ability to compensate for driver issues.
  10. headwhacker
    You are talking about the seal between the tip and the inner ear. But your general statement from what I understood suggest that limited bass extension is the result of intentional leakage, which open back headphone design I assume falls under this description.
  11. Niouke
    something I've read on facebook, the topic was to improve the sound in your car (waste of time if you ask me):

    "I listen to super chromium or metal cassette because with cassette I can sample the output of my high end DAC that sits at home AND I can easily swap cassettes while in my car to change what tune that I want to listen to. I use a Pioneer FH-P5000MP head unit in the car and a JVC TD-V66 at home to master the cassettes.'

    and it goes on:

    "When you want to copy something you don't want to make things worse. So by playing back a digital file in my cheap head unit inside of my car with a cheap nasty 1-bit DAC instead of the TDA1541 that I have at home in the DAC which sits at home I am worsening things, by going from a TDA1541 to a 1-bit DAC, so by copying the output of my TDA1541 at home to a cassette I can listen to the output of my TDA1541 in my car, which is, too big to be put into a car."
    Last edited: Sep 29, 2017
  12. Niouke
    I do have a question about one statement tho:
    "This is a recording of Billy Joel live in Yankee Stadium recorded off of a satellite feed in 1990, the source was completely analog the entire way from the USA as this was a live recording and satellite TVRO hadn't moved to digital encoding yet in 1990. Quite a journey for an analog signal, yet it made it intact!?"

    Was there really a time when end consumer satellite links were all analog?
  13. pinnahertz
    Absolutely. Satellite transmission followed the same development curve as pretty much everything, starting with analog, then as digital conversion and coding developed, digital became practical. Home TVRO systems (TV Receive Only) used huge dishes in the back yard, 4' to 10' in diameter, and with motorized positioning because you couldn't get every channel you wanted from one satellite. Positioning information was stored in the receiver. Polarization was also switchable. Initially there was no encryption even to premium channels, so it wasn't hard to bootleg HBO for example. Eventually they were all encrypted, then you could buy a bootleg de-encrypted receiver, Spy vs Spy, and so on until the move to digital, then the whole mess reset and started over.

    Even a lot of audio-only services were analog subcarriers on TV channels, which didn't work very well. So some audio only services like those used by NPR used a system called SCPC, Single Channel Per Carrier, the operation of which should be kind of obvious, carriers used FM on a transponder with no video, and were barely above the noise floor even with huge dishes and the best NLA of the time. The resulting audio noise floor was unacceptable, so NPR developed a companding noise reduction system akin to dbx except the ratio was 3:1. It sort of worked most of the time.
  14. Arpiben
    Yes there was a time were satellite carriers were dealing with more analog signals than nowadays.
    Using digital signals allow
    • use of complex digital modulations (8 PSK,32 APSK)
    • forward error correction techniques
    • use of digital video compression techniques
    • frequency reuse techniques
    • smaller dishes
    • etc
    With the same allocated bandwidth you carry n*10 times mores channels vs analog transmission.
    This is called bandwidth efficiency.
    Same happen with earth transmission networks (Microwave Transmission) were digital modulation are reaching 4096QAM at max.

    Dealing with latency in satellite transmission (journey) it is around 600ms when dealing with satellites with geostationery orbit (36000Km); 240 ms for the waves to travel up&down add the remaining ms for digital treatment.
  15. headdict
    I'm not the one who made that general statement, so I don't know exactly what was meant by that. I simply responded to your statement that "iSines are open back and doesn't require a tight seal from the tip".
254 255 256 257 258 259 260 261 262 263
265 266 267 268 269 270 271 272 273 274

Share This Page