Debate Thread: Scott's Hi-res Audio Test - Page 52 - AVS Forum
First ... 50  51  52 53  54  ... Last
Audio Theory, Setup, and Chat > Debate Thread: Scott's Hi-res Audio Test
stereoeditor's Avatar stereoeditor 05:00 PM 07-03-2014
Quote:
Originally Posted by Chu Gai View Post
That was generous and thoughtful of you John, seriously.
I hope others do likewise.

Quote:
Originally Posted by Chu Gai View Post
No central air (not such a big surprise in B'kln) but does the room have a window unit?
No, just a ceiling fan, which I turn off for serious listening in the summer. (Because it's a basement room, with the bottom half below ground level, it takes a long time to heat up.) We do have window units in the living room, kitchen and bedrooms in the other three floors, however.

John Atkinson
Editor, Stereophile

Randy Bessinger's Avatar Randy Bessinger 06:13 PM 07-03-2014
Quote:
Originally Posted by stereoeditor View Post
I have three spl meters, ranging from an antique RadioShack analog meter to the combination of the Studio Six Digital SPL meter app for the iPad, which is used with Studio Six Digital's iTestMic. (The Studio Six iTestMic is supplied with a correction file, generated by comparison with a B&K lab mike.) I also have an AudioControl SA3050A 1/3-octave analyzer, which has an spl function. These all systems agree within 3dB or so but I haven't checked them against my DRA Labs MLSSA system, which is used with a calibrated DPA 4006 mike but is housed elsewhere in the house.

Using the Studio Six app on my iPad and the iTestMic, the spl of the background noise in my dedicated basement listening room measures 40.5dB(C) and 28.6dB(A), both taken with the meter's ballistics set to Slow. To generate the attached spectrum of the noise in my room, I played a 1kHz tone at 95dB spl at the listening chair, measured with the Studio Six app & mike, and adjusted the preamp gain on an EMU404 USB ADC to give a digital level of -3dBFS. I then turned off the tone and captured the room tone for 30 seconds. Adding 98dB to each FFT bin allowed me to plot the spectrum with an spl vertical scale.

My room is very quiet, with the main source of noise distant traffic. (I don't have central AC.) However, it is not as quiet as some of Stereophile's reviewers' rooms, who live in rural locations. Acoustic treatment uses ASC Tube Traps and RPG Abfussors, and three of the four walls are lined with bookshelves and CD/LP racks. My goal was to make the room articulate and uncolored but not too dead. Rt60 is an even 250ms in the midrange and low treble and increasingly lower than that in the top two octaves.

Now that I have posted the spectrum of my room's background noise, I anticipate that other AVS forum members will do likewise.

John Atkinson
Editor, Stereophile
IF you go up and request that in the home theater construction forum I am sure you will get a number of responses. I just went to "The Savoy" home theater opening but don't remember the number or relevant numbers. It was very quiet (consulting by someone Amir knows well....but DIY by theater owner). Mine was measured by a member of the Erskine group and mine was just ok. I do have loud air conditioning but there is an off switch:-)

By the way, FWIW, an arbitrary distinction between home theater or multi channel and two channel stereo among almost all the young home theate/multi purpose room guys is a mistake. Many of them are avid two channel guys as well. Most just are looking for better sound for much less money than is considered the "high end". A good search would be of "Carp" (AVS name) of posts on this forum. You will find that he just purchased JTR Neosis 215's. Most of these people do NOT post here or in the Ultra High End sub-sections. Most are in their 20-30's and will travel all over the country to hear "meets" come and listen invites. They do their own blind speaker listening tests as well as amplifier tests. They are "doers" rather than readers and followers.
Audionut11's Avatar Audionut11 06:35 PM 07-03-2014
FWIW.

I live not far from a rural town, and it's the middle of the day here. Might do this again when I do my serious music listening (9pm+).
The spikes in the SPL graph are cars driving past. Not sure if I have the RTA set correctly.

dB A weighted, is a pretty consistent 34.9.
Attached: spl.png (131.0 KB)  rta.png (108.0 KB) 
krabapple's Avatar krabapple 06:35 PM 07-03-2014
Quote:
Originally Posted by arnyk View Post
The words bias controlled are wholly gratuitous in this context.

When a piece of equipment turns itself off and flashes a red light at you, no listening test at all should be needed.

The original (ca. 1975) statement of the requirement for bias-controlled testing was:

"Bias controlled listening tests are required for all situations involving subtle differences."
And here is the thing: differences that from a priori considerations should be *at best* subtle, if not wholly imaginary, are often trumpeted as 'huge' by 'subjectivists' (by which I mean those who put 'sighted' perception of audible difference on par with blind, or even *superior to it*. I include in their ranks a certain 'stereoeditor' battling tediously with you on this thread). Does it it make sense to give *them* the benefit of doubt?

Science has long since made its decision on that.
krabapple's Avatar krabapple 06:42 PM 07-03-2014
Quote:
Originally Posted by CharlesJ View Post
When was the education such that people didn't have a problem properly interpreting their senses?

Human gullibility is very powerful, innate part of us; few can have it under control and even well educated even in the sciences are not immune.

Hard work.
Controls are mandatory in scientific work because scientists understand that they are not immune.

A proper education in the scientific method includes this. If I learned it, others must have too.
stereoeditor's Avatar stereoeditor 06:44 PM 07-03-2014
Quote:
Originally Posted by krabapple View Post
I include in their ranks a certain 'stereoeditor' battling tediously with you on this thread).
Why the personal comment?

John Atkinson
Editor, Stereophile
krabapple's Avatar krabapple 08:32 PM 07-03-2014
Quote:
Originally Posted by stereoeditor View Post
Why the personal comment?

John Atkinson
Editor, Stereophile
You mean calling your eternal and quite personal argument with Arny 'tedious'?

Or mentioning you at all while citing the cohort that believes 'subjective' impressions of audible difference are on par with, or superior to, controlled comparisons?

You're the editor of Stereophile. Your magazine is one of the two most popular and best known 'audiophile' journals, and it has played a sure and steady role in promulgating the 'subjectivist' stance in the hobby, for decades, certainly as long as you've run it. So, own it.
amirm's Avatar amirm 09:29 PM 07-03-2014
Quote:
Originally Posted by krabapple View Post
You mean calling your eternal and quite personal argument with Arny 'tedious'?

Or mentioning you at all while citing the cohort that believes 'subjective' impressions of audible difference are on par with, or superior to, controlled comparisons?

You're the editor of Stereophile. Your magazine is one of the two most popular and best known 'audiophile' journals, and it has played a sure and steady role in promulgating the 'subjectivist' stance in the hobby, for decades, certainly as long as you've run it. So, own it.
Stereophile is the largest repository of objective measurements and data anywhere. Our own Arny quotes graphs from it left and right. And they provide all of this data for free for "us objectivists" to use to battle the subjectivists.

That you suffer from emotional angst in what they are doing is symptom of not being interested in learning. I hope you are able to own this. Lacking that, please make sure your next post is technical and something new.
spkr's Avatar spkr 10:49 PM 07-03-2014
Quote:
Originally Posted by amirm View Post
Stereophile is the largest repository of objective measurements and data anywhere.
As well as subjective reviews.
RayDunzl's Avatar RayDunzl 10:57 PM 07-03-2014
(spkr attacking me about something I perceived with my ear/brain)
Quote:
Originally Posted by spkr View Post
With microphone. It may be inconvenient for some but not for those with experience and the equipments already set up.
I suppose you haven't tried to measure it.
For the benefit of anyone else who thinks I may not be interested in measurements, here is my evening noise floor.

8 minutes just after midnight.
PC fan needs to die,other than "self noise" of the test gear. It's the only thing I hear. I moved the mic far away from the PC after the test, level dropped only 1db. So, I will assume the readings are mostly just the cheezy test equipment measuring itself.

Test Condition:


1. Test noise from REW - microphone is the monitor output of a Tascam DR07 MKII handheld (on tripod at sweet spot)
2. Antique Analog RatShack sound level meter
3. Adjust noise to 70db on the meter, C-weighted
4. Calibrate REW to read 70db C-weighted


REW SPL - when the AC finally came on it rose to 40-41





Mostly electrical noise - PC sound card spikes up high, a little 60hz - 25 foot cable to Tascam



The noise peak - I flicked my bic.




edit: With the tascam turned off, cable still attached to it and the PC onboard soundcard, REW shows 30.2dB(C), and the spectrum is a little cleaner (60hz is gone, and some of the HF hash is gone/reduced) but barely (10db or less across the band) any different.
RayDunzl's Avatar RayDunzl 11:08 PM 07-03-2014
Quote:
Originally Posted by RobertR View Post
Really? Lower noise causes one to perceive a "wider soundstage"?
Now that I have had a while to think about it, yes, lower noise does assist in the perception of a "wider soundstage".

It's late night when things are very quiet that the imaging really pops.

Does it matter if the noise is in the recording or in the equipment or in the room? That's a good question, I'm glad I thought of it. I may have to play with that thought some, and see what I can perceive.
spkr's Avatar spkr 11:31 PM 07-03-2014
Quote:
Originally Posted by RayDunzl View Post
(spkr attacking me
That's just your perception. The pitfall of personal perception is that it's all over the place. That's why the aid of measurement is important in audio field.
Quote:
Originally Posted by RayDunzl View Post
about something I perceived with my ear/brain)
If your perception is triggered by the actual soundwave hitting your eardrums, the audible changes they cause are certainly measurable.
RayDunzl's Avatar RayDunzl 12:02 AM 07-04-2014
Would you care to give an example?
RobertR's Avatar RobertR 12:02 AM 07-04-2014
Quote:
Originally Posted by RayDunzl View Post
Now that I have had a while to think about it, yes, lower noise does assist in the perception of a "wider soundstage".

It's late night when things are very quiet that the imaging really pops.

Does it matter if the noise is in the recording or in the equipment or in the room?
It matters if you're trying to differentiate between different equipment, which is the whole point.
RayDunzl's Avatar RayDunzl 12:29 AM 07-04-2014
Sorry, I was not differentiating between different equipment, and it is not the whole point to me.

I was responding to your statement "Really? Lower noise causes one to perceive a "wider soundstage"?" and that alone.
Kees de Visser's Avatar Kees de Visser 12:48 AM 07-04-2014
Quote:
Originally Posted by Randy Bessinger View Post
I am not sure I understand. Wasn't Kees just using the test as a diagnostic, not that he could pass an ABX test of the two files? Seems to me they were just making sure that the two files were correctly leveled. Otherwise, someone could complain later.. Kees (or ISA), correct me if I am wrong.
I have just tested the validity of the supplied source material. Similar to making sure the med lab instruments are sterile Sorry if this wasn't clear.
RayDunzl's Avatar RayDunzl 02:20 AM 07-04-2014
I just ran across this abstract after reading Scott's fine new article on the High-Resolution Audio Listening Event at CE Week 2014. The phrase "unfettered dynamics" caught my eye.


"A dynamic range of 118 dB is determined necessary for subjective noise-free reproduction of music in a dithered digital audio recorder. Maximum peak sound levels in music are compared to the minimum discernible level of white noise in a quiet listening situation. Microphone noise limitations, monitoring loudspeaker capabilities, and performance environment noise levels are also considered."

Author: Fielder, Louis D.
Affiliation: Ampex Corporation, Redwood City, CA
AES Convention:69 (May 1981) Paper Number:1772 Import into BibTeX
Publication Date:May 1, 1981

Maybe someone could someone post some pertinent part of it, as part of this 16 vs 24 discussion?



(apologies if it has been covered, I can't debate-read as well as perhaps I should)
arnyk's Avatar arnyk 05:21 AM 07-04-2014
Quote:
Originally Posted by stereoeditor View Post
I know you addressed this to Amir, but I just posted the spl and spectrum of the background noise in my listening room. The spl meter (iPad with $199 Studio Six iTestMic) and mike preamp-A/D converter (an E-MU 0404 for which I paid $180) were inexpensive and while I used a pricey Earthworks QTC-40 omni mike for the spectral analysis, there is an excellent Behringer omni mike, the ECM8000, available for just $60. I performed the FFT analysis with Adobe Audition, which I know you already own.

Look forward to seeing a similar analysis for your room, Mr. Krueger.
Nice graph that may raise more questions than it answers, What is the SPL, A weighted and C weighted? What conditions?

I have the ideal tools for doing this but I have to find it all! Most elusive seems to be my Microtrack...

The easy part is SPL which is about 43 dB at the quietest time in the middle of the night, and more like 52 dB during the day, providing nothing special is happening. This is with our house in summer configuration which has all the windows open so that we can enjoy our delicious Michigan late spring and early summer. In the winter with all windows closed I would expect even lower measurements than the summer best 43 dB, at the worst.
stereoeditor's Avatar stereoeditor 05:30 AM 07-04-2014
Quote:
Originally Posted by krabapple View Post
Quote:
Originally Posted by stereoeditor
Quote:
Originally Posted by krabapple
I include in their ranks a certain 'stereoeditor' battling tediously with you on this thread).
Why the personal comment?
You mean calling your eternal and quite personal argument with Arny 'tedious'?
Yes. That's the literal meaning of the quoted text, other than your interjection of "quite personal," which is, of course, a subjective judgment. I fail to see, therefore, how your personal comment was either relevant or appropriate. So it goes.

John Atkinson
Editor, Stereophile
arnyk's Avatar arnyk 05:38 AM 07-04-2014
Quote:
Originally Posted by RayDunzl View Post
I just ran across this abstract after reading Scott's fine new article on the High-Resolution Audio Listening Event at CE Week 2014. The phrase "unfettered dynamics" caught my eye.


"A dynamic range of 118 dB is determined necessary for subjective noise-free reproduction of music in a dithered digital audio recorder. Maximum peak sound levels in music are compared to the minimum discernible level of white noise in a quiet listening situation. Microphone noise limitations, monitoring loudspeaker capabilities, and performance environment noise levels are also considered."

Author: Fielder, Louis D.
Affiliation: Ampex Corporation, Redwood City, CA
AES Convention:69 (May 1981) Paper Number:1772 Import into BibTeX
Publication Date:May 1, 1981

Maybe someone could someone post some pertinent part of it, as part of this 16 vs 24 discussion?



(apologies if it has been covered, I can't debate-read as well as perhaps I should)
I have a PDF of the entire paper. PM me if interested. The paper is highly optimistic, an excluded middle argument for extreme amounts of dynamic range if there ever was one. It ignores many significant influences such as the fact that listening to music at 118 dB SPL precludes hearing quiet sounds at 0 dB for a quite a while later until the ears re-adjust. And that 118 dB SPL number is extreme. The low SPL numbers lowball real world situations. I think they may even ignore the crowd noise at live concerts and instead concentrate on data from the quietest venues around when totally deserted and in the middle of the night.

The sensitivity of the ears seems to start decreasing slightly when SPLs reach 85 or 90 dB for any period of time. Higher SPLs for longer periods, more desensitization.

Reality is that few commercial recordings regardless of medium have more than 75 dB dynamic range (68 dB is more typical for uncompressed works), and none that I can find exceed 89 dB (and I wonder if there was gain riding during their production). Yet, I find few complaints about objectionable hiss or other noises during the quiet passages.
stereoeditor's Avatar stereoeditor 05:41 AM 07-04-2014
Quote:
Originally Posted by arnyk View Post
Nice graph that may raise more questions than it answers, What is the SPL, A weighted and C weighted?
Both were mentioned in the text accompanying the graph, but to repeat: 40.5dB(C) and 28.6dB(A), both taken with the meter's ballistics set to Slow.


Quote:
Originally Posted by arnyk View Post
What conditions?
Measurement in afternoon; windows closed (they don't actually open); mike at listening chair; ceiling fan off (as I explained, it is always turned off for serious listening). My laptop was on, of course, and that's most likely the source of the low-level discrete tones at high frequencies in the spectrum.

Quote:
Originally Posted by arnyk View Post
The easy part is SPL which is about 43 dB at the quietest time in the middle of the night, and more like 52 dB during the day.
A weighted or C weighted? Meter ballistics?

John Atkinson
Editor, Stereophile
arnyk's Avatar arnyk 05:47 AM 07-04-2014
Quote:
Originally Posted by Randy Bessinger View Post
I am not sure I understand. Wasn't Kees just using the test as a diagnostic, not that he could pass an ABX test of the two files? Seems to me they were just making sure that the two files were correctly leveled. Otherwise, someone could complain later.. Kees (or ISA), correct me if I am wrong.
Begging the question, how do we know that the files are properly leveled?

One way is to look at accepted custom among experienced experimenters which says that +/- 0.1 dB (0.2 dB range) suffices.

Another way would to be examination of the Clark JAES ABX paper, much if not all of which stands scrutiny from all but the usual golden ears to this day.

If you want to go the DIY route then the best way would probably to do a number of ABX comparisons. Good practice is to start with files that are highly mismatched and easy to detect, and as a training exercise decrease the amount of mismatch in reasonable steps until reliable detection ceases.

Then cut the first mismatch that eluded detection in say half, to provide a reasonable safety margin.

Safety margins are important - no engineer wants to build a bridge that collapses when one more person than its usual maximum load stands on it.

Seems like common sense to me. This is the pattern I followed with the recent jitter tests. Comments?
arnyk's Avatar arnyk 05:54 AM 07-04-2014
Quote:
Originally Posted by stereoeditor View Post
Both were mentioned in the text accompanying the graph,
Not on the graph itself, which is all I downloaded.

Quote:
Originally Posted by stereoeditor View Post
40.5dB(C) and 28.6dB(A), both taken with the meter's ballistics set to Slow.

Measurement in afternoon; windows closed (they don't actually open); mike at listening chair; ceiling fan off (as I explained, it is always turned off for serious listening). My laptop was on, of course, and that's most likely the source of the low-level discrete tones at high frequencies in the spectrum.


A weighted or C weighted? Meter ballistics?
My numbers are C weighted slow in both cases. Your numbers are probably pretty close to what I'd expect to see in the winter under similar TOD conditions.

When I find the Microtrack I'll redo all of the measurements at both times of day. Nice things about the Microtrack - not only highly portable but no fan.
stereoeditor's Avatar stereoeditor 05:57 AM 07-04-2014
Quote:
Originally Posted by arnyk View Post
"A dynamic range of 118 dB is determined necessary for subjective noise-free reproduction of music in a dithered digital audio recorder. Maximum peak sound levels in music are compared to the minimum discernible level of white noise in a quiet listening situation. Microphone noise limitations, monitoring loudspeaker capabilities, and performance environment noise levels are also considered." - Louis D. Fielder,

...the fact that listening to music at 118 dB SPL precludes hearing quiet sounds at 0 dB for a quite a while later until the ears re-adjust. And that 118 dB SPL number is extreme.
At the risk of my comment again being referred to as "tedious," a dynamic range of 118dB is not equivalent to a continuous SPL of 118dB, the latter being what I inferred from Mr. Krueger's terminology. The abstract to Louis Fielder's AES paper, which I quoted above, specifically refers to peak levels.

John Atkinson
Editor, Stereophile
imagic's Avatar imagic 06:05 AM 07-04-2014
Quote:
Originally Posted by Randy Bessinger View Post
IF you go up and request that in the home theater construction forum I am sure you will get a number of responses. I just went to "The Savoy" home theater opening but don't remember the number or relevant numbers. It was very quiet (consulting by someone Amir knows well....but DIY by theater owner). Mine was measured by a member of the Erskine group and mine was just ok. I do have loud air conditioning but there is an off switch:-)

By the way, FWIW, an arbitrary distinction between home theater or multi channel and two channel stereo among almost all the young home theate/multi purpose room guys is a mistake. Many of them are avid two channel guys as well. Most just are looking for better sound for much less money than is considered the "high end". A good search would be of "Carp" (AVS name) of posts on this forum. You will find that he just purchased JTR Neosis 215's. Most of these people do NOT post here or in the Ultra High End sub-sections. Most are in their 20-30's and will travel all over the country to hear "meets" come and listen invites. They do their own blind speaker listening tests as well as amplifier tests. They are "doers" rather than readers and followers.
And these are the guys who truly turned me on to what's possible with home audio! They are the reason I can't take the whole "high end" argument seriously anymore. The AVS guys are the reason I wrote my "Is High End Audio Obsolete?" piece. They are the reason I built my own subwoofers. They are the reason I chuckle whenever my friend with $300,000 in Krell and Wilson brags about his system's capabilities—and he still won't listen to anything I say... evidently the brands and the money he spent help him feel superior in some way. One of these days I'll bring him to a proper AVS GTG and I'll hand him a copy of "Good Grief."

Frankly, I hold back (in relative terms) in my criticism of high-end only because most people have not experienced a system like the ones you find at an AVS GTG. When I went to an AVS GTG and the New York Audio Show back-to-back, that was the last straw. I am sooooo glad I got away from "high end" and into "high performance" audio. I have AVS Forum to thank for that.
arnyk's Avatar arnyk 06:16 AM 07-04-2014
Quote:
Originally Posted by krabapple View Post
And here is the thing: differences that from a priori considerations should be *at best* subtle, if not wholly imaginary, are often trumpeted as 'huge' by 'subjectivists' (by which I mean those who put 'sighted' perception of audible difference on par with blind, or even *superior to it*.
Absolutely true. I still remember a certain SZ (Who JA is also familiar with) who in the late 90s regaled us on RAO with "Mind blowing differences" due to different power amplifiers. This led to the "Sunshine Stereo Trials" proctored by the late TN. Needless to say those "Mind blowing differences" became elusive in listening tests with any reasonable amount of experimental controls.

IME even actual audible differences are often far less than mind blowing. What was mind blowing about the listening tests that were printed up in HFN was that such a highly regarded product as the ARC ST120 was so easily detected based on such a gross artifact. There have been some unpublicized power amp comparisons where audible differences were found, and at the end everybody agreed that even though there was a difference, it was not worth even looking across the street to see, let alone spend the big bucks for.

I would go so far as to say that *any* account of "Mind blowing differences" among what should be good DACs and power amps is highly suspect of being due to a poor listening test just because of the nature of the perception of the difference. Experience says that if there is an audible difference, its probably less than mind blowing.

Quote:
Originally Posted by krabapple View Post
I include in their ranks a certain 'stereoeditor' battling tediously with you on this thread).
As the saying goes, he has a dog in this fight. ;-)

Many mornings I wake up and ask myself whether its really worth the trouble to arm wrestle with the usual suspects yet another day. I'll get my pension checks regardless. Typing away on the web keeps me away from the best audio system I've ever owned and the best music collection I've ever had. My wife recently retired and I married her because I loved her company and now I'm squandering it?

Quote:
Originally Posted by krabapple View Post
Does it it make sense to give *them* the benefit of doubt?
I gave them all that back in 1975 and look what they did with it.

Quote:
Originally Posted by krabapple View Post
Science has long since made its decision on that.
The illusion that has not happened is their last frontier.

We own it.
arnyk's Avatar arnyk 06:28 AM 07-04-2014
Quote:
Originally Posted by stereoeditor View Post
At the risk of my comment again being referred to as "tedious," a dynamic range of 118dB is not equivalent to a continuous SPL of 118dB, the latter being what I inferred from Mr. Krueger's terminology
Well John, how do we break 118 dB dynamic range down into practical terms? As it stands, its just a number with no SPL reference.

If the reasonable reference is 0 dB SPL or more then 118 dB dynamic range equates to 118 dB SPL or more and I might take that as a win. ;-)

Due to apparent ambiguities in Fielder's paper, I find that the statistics of the sounds he used to establish the 118 dB dynamic range number seem ambiguous.

I'm perfectly willing to have a Fielder advocate enlighten me about things like crest factor, power bandwidth, spectral contents, etc.


Quote:
Originally Posted by stereoeditor View Post
The abstract to Louis Fielder's AES paper, which I quoted above, specifically refers to peak levels.
Whatever that means. I mean I know exactly what it means for an arbitrary electrical signal but I also know better than to claim that it is a psychoacoustically meaningful specification of a signal.

For one thing I know from ABXing that the ear blissfully bypasses a certain amount of clipping.
amirm's Avatar amirm 08:32 AM 07-04-2014
Quote:
Originally Posted by isa View Post
The Replaygain filter curve shown in Debate Thread: Scott's Hi-res Audio Test is no longer in use within foobar2000. foobar2000 developers assert Replaygain as currently implemented in foobar2000 uses the EBU R128 specs, which means it uses the following curve that does not truncate high frequencies
Ah, thanks for the correction. You are right of course.

Quote:
Originally Posted by isa

source: https://tech.ebu.ch/docs/tech/tech3343.pdf
If anyone has information that the foobar2000 developers didn't do what they claim to have done, please share your sources or analyses.

edit: while the image above stops at around 20k for display purposes, I see no high frequency cutoff in the underlying formulas of the filters that are used to construct that curve.
Foobar developers did not implement this themselves from my research online. They used the open source libebur128 library. Here is the computation of convolution kernel/filter coefficients in libebur128 (the b[] array):

Code:
double f0 = 1681.974450955533;
  double G  =    3.999843853973347;
  double Q  =    0.7071752369554196;

  double K  = tan(M_PI * f0 / (double) st->samplerate);
  double Vh = pow(10.0, G / 20.0);
  double Vb = pow(Vh, 0.4996667741545416);

  double pb[3] = {0.0,  0.0, 0.0};
  double pa[3] = {1.0,  0.0, 0.0};
  double rb[3] = {1.0, -2.0, 1.0};
  double ra[3] = {1.0,  0.0, 0.0};

  double a0 =      1.0 + K / Q + K * K      ;
  pb[0] =     (Vh + Vb * K / Q + K * K) / a0;
  pb[1] =           2.0 * (K * K -  Vh) / a0;
  pb[2] =     (Vh - Vb * K / Q + K * K) / a0;
  pa[1] =           2.0 * (K * K - 1.0) / a0;
  pa[2] =         (1.0 - K / Q + K * K) / a0;

   f0 = 38.13547087602444;
  Q  =  0.5003270373238773;
  K  = tan(M_PI * f0 / (double) st->samplerate);

  ra[1] =   2.0 * (K * K - 1.0) / (1.0 + K / Q + K * K);
  ra[2] = (1.0 - K / Q + K * K) / (1.0 + K / Q + K * K);


  st->d->b[0] = pb[0] * rb[0];
  st->d->b[1] = pb[0] * rb[1] + pb[1] * rb[0];
  st->d->b[2] = pb[0] * rb[2] + pb[1] * rb[1] + pb[2] * rb[0];
  st->d->b[3] = pb[1] * rb[2] + pb[2] * rb[1];
  st->d->b[4] = pb[2] * rb[2];
The filter response graph you post by the way is from BS.1770. EBU 128 is just a set of recommendations on top of BS.1770. BS.1770 recommends 4X upsampling for 48 Khz and 2X for 96 Khz. I can't quite tell how resampling is done in the above library. Sample rate is used to change the corner frequency (K) but that is it. There is code to call a resampler but it is commented out.

The person who wrote the above library believed in fewer comments in his code, the better which makes it a pain to read through his code quickly and grok what he is doing. I am too lazy to be a human computer and try to figure it all out by reading the code .

Since "ReplayGain" is based on BS.1770, and I have that computation in my Adobe Audition, I sampled down and up one of the high-res version of Scott's file. The LUFS loudness value was accurate to two decimal places and did not change in this back to back conversion. More testing is needed but it seemed to me that it did not incorporate the high-frequency spectrum.

Do you have more data on this? And thanks again for the correction .
spkr's Avatar spkr 08:40 AM 07-04-2014
Quote:
Originally Posted by RayDunzl View Post
Would you care to give an example?
An example of what? Would you mind quoting the specific sentence when asking such question?
amirm's Avatar amirm 09:33 AM 07-04-2014
Quote:
Originally Posted by spkr View Post
If your perception is triggered by the actual soundwave hitting your eardrums, the audible changes they cause are certainly measurable.
That is like saying world peace is achievable. There is a big difference between claiming that and ability to get it done.

Our audio measurements are ancient relics from decades back. Dealing with today's audio systems requires far more sophistication which lacks in measurements people perform. Take the recent discussion of room noise. As I explained the SPL meter is useless at best, misleading at worst. You can measure the frequency response but then you have to interpret that graph relative to hearing system of many people. The measurement was a simple frequency response devoid of psychoacoustics.

Take a system where high frequencies are merged between channels only during sharp transitions. There is no practiced measurement system that tells you what is going on there.

Take audio compression. We don't use a single measurement to assess their fidelity.

Take Arny's "jitter" tests and eliminate the starting tone. Now you tell me how you would determine level of jitter in any of them from measurements. How would you distinguish music from music+jitter?

So let's not throw out talking points like "you can measure it." Yes it is possible. But possibility doesn't mean availability. It is possible to travel to outside of the solar system. But no one has done so.

At some point we have to declare these tired arguments for what they are: tired arguments. Who hasn't heard "oh, you can measure it." Come back when you have a novel way to measure what is being talked about.
First ... 50  51  52 53  54  ... Last

Up
Mobile  Desktop