View RSS Feed

mitchco

Hear music the way it was intended to be reproduced - part 4

Rating: 2 votes, 5.00 average.
This is Part 4 of a series on a quest to hear music the way it was intended to be reproduced. In the last 3 posts, we have “voiced” and calibrated our speakers to an equilateral triangle, took some frequency response measurements, analyzed the results, and introduced digital room correction (DRC). Let’s look at the frequency response measurement results and DRC in more detail. Is DRC ready for prime time? I think this post will show conclusively, yes, DRC is ready for audiophiles to take full advantage of their sound system investments. Once you hear correct timbre, you won’t go back ;-)

When I say DRC, I mean the subject area: http://en.wikipedia.org/wiki/Digital_room_correction and not the software with the same name: http://drc-fir.sourceforge.net/ I have not used this DRC software and can only provide you with my experiences using Audiolense.

Here is the frequency response of my sound system measured at the listening position:



Here is the frequency response with DRC enabled, using the B&K house curve as the target, again at the listening position:



How was this DRC accomplished? Within Audiolense, you click on generate correction filters, which produces the inverse of the measured frequency response like this:



You save the filter to a file location and then load it in a Convolver like the one hosted in JRiver Media Center:



And within a few minutes, you are listening to music the way it was intended to be reproduced. More on this later.

Now let’s zoom in on these graphs so you can see more detail. I will use the same relative vertical and horizontal scales so that the two graphs can be compared.

Without DRC at the listening position:



With DRC (and the B&K house curve applied) at the listening position:



Let’s analyze this. Given the graphs presented, I would conclude that DRC works very well. It's quite the controlled difference between the before and after with DRC, especially the tight variation tolerance. After 200Hz, the variance is +- 2.5db. Tighter tolerance than the manufacturers anechoic chamber specs.

Here is what it means. Most speaker manufactures will produce a frequency response curve (plus speaker sensitivity) @ 1 watt @ 1 meter in an anechoic chamber. The tech spec usually includes a variance limit on the frequency response, like +- 3db across the measured frequency range. Note that the anechoic chamber is designed to eliminate the room effects on the measurement. Having been in one, it is an interesting experience, completely void of relected sound.

Let’s take the popular B&W CM8 and look at its specified frequency response. From their online manual, http://www.bowers-wilkins.com/Downlo...info_sheet.pdf, the frequency response measures 69Hz to 22Khz +- 3db variance. My speakers, which are a tech modernization of the Klipsch Cornwall, but custom designed and built by Bob Crites, called Cornscala Type C http://www.critesspeakers.com/cornscala-style-c.html They will have a similar frequency response to the Cornwall III’s of 34hz to 20Khz +- 3db.

My point is that while my speakers measure, 34Hz to 20Khz with a +- 3db variation in an anechoic chamber, the moment that I put them in a real listening room, of any sort, all bets are off. Look at the frequency response variations of my listening room again. Note the maximum amplitude deviation.

I get +- 12 db from about 34Hz to almost 20Khz. Having measured several studio control rooms and critical listening environments, this looks typical. In fact, regardless of your audio electronics and speakers, you are likely to get similar measurements in your own listening room – it’s all a function of room modes and there is no escape. I could certainly improve mine by moving the speakers/listening position around a bit more and adding acoustical treatments to the room. Ultimately, I could build a seperate critical listening room with more favorable "golden room ratios", but that isn't in the cards for me at this time. Side note, to be sure, if you ever have the opportunity to build a room, get the golden rule ratios as it does make a fundamental difference in the low end.

If I look at the DRC “calibrated” frequency response above and not taking into account the slope of the B&K house curve, I get +-3 db from about 34Hz to almost 20Khz. Now that is about the same frequency response specification I get from the manufacturer, when they measure in an anechoic chamber.

Effectively, what this means is the DRC is not only doing its job by eliminating or minimizing the room acoustics, it is also applying (i.e. calibrated by) the B&K house curve which renders the right tonal quality (i.e. timbre) at the listening position. This is even without the most basic of acoustic room treatments. I could use a carpet on the hardwood floor in front of the speakers. Note how closely the left and right curves match each other. This means that we get a solid dead center phantom image produced by the speakers as any amplitude imbalance of the electronics/speakers/room interface has been calibrated by the DRC.

I have been using Audiolense (and the DRC filter it produces) in my audiophile system for about 6 months. I have not had any issues with the digital filters or any other critical listening artifacts arising from their use.

I look at DRC as a must have to fully realize your audio system investment. Every attention to detail and calibration throughout the audio chain will pay off in the end. However, given that modern day electronics can have frequency responses from 10Hz to 100Khz with +- .25db variations, the speaker to room interface is the biggest variation, by far, on the quality of sound (read: timbre) than any other component in the audio chain.

You are not locked into the listening position using frequency DRC. You can walk anywhere around your listening room and notice the major improvement in sound quality. As mentioned earlier in my first post, my wife commented on how the recorded piano I was listening to still sounded real (i.e. correct tone quality) when she was out in the attached garage. Correct timbre, once you hear it, you won’t go back.

Or at least that has been my case. I am still blown away listening to my studio mixes from many years ago that I produced in a multi-million dollar recording facility and hearing it sound “identical” in my home. I would suggest that with DRC, and the B&K house curve, you hear music reproduced as close as possible to exactly what the mastering/mixing engineer and producer wanted you to hear, with the proper timbre. With the advent of 96/24 resolution recordings, you are hearing as close to the master tape as possible. It is great to be a computer audiophile!

We can get even more sonic improvements. In the case of Audiolense, we can achieve time domain correction in addition to frequency correction. What does this mean? We will look at this in my next post.

Happy listening!

Mitch

Updated 05-09-2012 at 01:52 AM by mitchco

Categories
Personal Blogs

Comments

  1. mbain's Avatar
    As a Mac user are there alternatives to Audiolense as a digital room correction software package? My understanding is that Audiolense is only PC friendly.
  2. mitchco's Avatar
    Hi mbain, yes Audiolense is PC only. Sorry, I don't know what the alternatives are on the Mac. Maybe someone else on the forum knows...
  3. Bradc's Avatar
    What does the corrected response look like if you move the microphone 3" in any direction?
  4. mitchco's Avatar
    Hey Bradc,



    Actually there is little variation when I move the mic around the listening area (couch) by several feet. You are not "locked into" a sweet spot of inches. The measurement mic is omnidirectional: http://en.wikipedia.org/wiki/Microphone#Omnidirectional







    Audiolense has multi-seat correction. The above example is from the Audiolense help file. You can download the help file from: http://www.juicehifi.com/index.html as it goes into a more detailed explanation of how it works.



    Hope that helps.
    Updated 05-09-2012 at 10:57 AM by mitchco
  5. Bradc's Avatar
    Hi

    I use acourate myself and agree that DRC is quite impressive, especially with the new pre ringing compensation.

    My point about moving the mic is that the plotted response does not look that well corrected, but it still sounds really good. For frequencies above approx 1kHz spatially averaged measurements are required as it outside of the modal region. Earl Geddes has a lot to say on this topic, but I can't find the references at the moment.



    With acourate, the sweet spot that I get is about a seat width wide.

    Do you get good stereo imaging from wider seating positions than that with your setup/audiolense?

    Brad
  6. mitchco's Avatar
    Yes, pinpoint 3D imaging from really anywhere in the room, with a caveat. It is the difference between using just frequency correction, or full correction in both the frequency and time domain, called True Time Domain (TTD) in Audiolense.



    Using just frequency correction and tuning the target response to the B&K house curve at the listening position:







    Now the room has that "tone quality" or "tonal shape" and to me and others, has the most natural timbre from a frequency response perspective. The tonal balance sounds exactly like my Sennheiser PXC 450 headphones. The "tone quality" changes very little and predictably as I wander the room.



    The stereo image is perfectly stable. It sounds like how you would think it should sound if you walked around a visual projection of your speakers to listening position as a 3D equalateral triangle.



    The more I used a laser distance measure to align the speakers as symetrically as possible to the back wall and form as close to a 1/4" tolerance equalateral triangle made a huge difference in not only pinpoint imaging, but in tone quality too. In addition, as you can see by the graphs in the article how close the left and right channel frequency responses are "matched" in amplitude. The phantom center image is a good as headphones.



    When you engage Audiolense TTD, the software corrects in the time domain, and works well across the width of a 3 person couch. If you got up and wandered around the room, you hear the "out of sync" timing.



    As you move closer to the speakers, you pick up on preringing or the sound of the transient just coming before it peaks. It sounds like a kick drum played backwards. Move behind the couch and you hear the late arrival of out of sync sound, so the couch is the listening area.



    But the payoff is it sounds incredible. I am amazed at the state of the art in DRC. The transient impact arrives at your ears at the same time (time aligned). So the sound is more impactful and in perfect 3D audio.



    Audiolense generated filters also clean up early reflections so you can tune how clear (from nothing to "headphones") you want the sound.



    And fiddling with the bass time domain window, I was able to extend to 20Hz yet sound really tight in the room. It feels like Bernt designed it this way. I am really impressed with the quality of filters generated by his software.



    I took several measurements with different mic directions (straight up, on, and 45 degree angle) at various positions on the couch. It was a wash, basically the same frequency response once averaged. It also sounded consistent across the couch. So, I have not tried the multiseat correction yet.



    Hope that helps,



    Mitch
    Updated 05-09-2012 at 10:57 AM by mitchco