View RSS Feed


16/44 vs 24/192 Experiment

Rating: 5 votes, 5.00 average.
Updated with more info on Audio DiffMaker, plus ABX listening tests.

Lots of discussion around this article: 24/192 Music Downloads...and why they make no sense

I decided to run a science experiment using Audio DiffMaker to compare 16/44 to 24/192 format of the same master from Soundkeeper Recordings:

I have used Audio DiffMaker before to compare FLAC vs WAV and comparing two bit-perfect music players on my computer audio playback system.

Here is the result of my 16/44 vs 24/192 experiment.

First a refresher on how Audio DiffMaker works:

There are also a two papers, and The help file that comes with the program is very well documented and goes into much more detail.

Updated - I wanted to provide more with respect to how Audio DiffMaker works and why it is an important state of the art measurement tool in any Audiophiles arsenal.

Audio DiffMaker’s Differencing Process

Excerpt from the DiffMaker Help file on how the differencing process works:

While it may not be possible to show whether alteration is having effects directly on the listener, it is possible to determine whether an audio signal has been changed.
The existence of any changes to a digital recording of an audio signal can be detected by the simple process of subtraction, performed on a sample-by-sample basis. If each audio sample is the same, then subtracting one from the other leaves nothing (zero signal).

A recorded copy of the original signal (called the "Reference") can be mathematically subtracted from a recorded copy of the possibly changed signal (called the "Compared" signal). This results in a "Difference" signal recording that can be evaluated by ear or other analysis.
If the resulting Difference signal, when played as audio, is effectively silence or at least is not perceivable to a listener when played at levels in which it would occur when it was part of the "Compared" signal, then the investigator can with good confidence conclude that the change has made no audible difference.

The problems and operational, perceptual, or psychological complications about listening for whether sound is being changed are greatly reduced by transforming the task into the much simpler issue of listening for anything significant at all. The evaluation of the result is done by ear, and the user doesn't need to question hearing ability to use the tool. Audio DiffMaker test, encourages you to still "trust your ears".

Audio DiffMaker is a state of the art differencing tool that automates this workflow from 5 years ago: One of the reasons it is state of the art is because the software can differentiate time differences in decimal places in the parts per million (ppm): “The sample rates or speeds of player decks and soundcards are constantly drifting, if only by very small amounts. But even as little a change in sample rate as 0.01ppm (one hundredth of a part per million) can cause two otherwise identical files to leave difference sound after subtracting.”

In order to compare the two formats, I had to up sample the 16/44 to 24/192. I used to perform the sample rate conversion:

I used the default settings. Then I used Audacity to edit the waveforms so I am just looking at the first 40 seconds of each waveform.

Then it is a matter of loading the two waveforms into Audio DiffMaker and extracting the difference.

According to DiffMaker, the difference file is -94 dB. I opened up the difference file in Audacity and here is what is left over:

Something definitely there. Here is the frequency analysis:

I have also included the difference file as an attachment to this post. Given that the majority of content is 20KHz and above, I can’t hear anything on the difference file.

Note that this is one data point. I have used Audio DiffMaker for a while now and here is one tip that will help you get consistent results if you decide to try it out.

This is the output status window from the DiffMaker progam as it is running. Note the arrow. It says that the sample rate error is low enough not to require adjustment. If the sample rate error is too high, there will be a notification as such on this line, then the program tries to automatically align the tracks. However, there seems to be a bug in the program, as noted in one of my other posts, so the track alignment does not seem to work or work very well. Therefore, I am unable to get consistent results.

If you look in the status window and see that your comparison requires sample rate adjustment, then here is what you can do. Open up the waveforms in your favorite digital audio editor and ensure that the both waveforms “start” at exactly the same time. That’s the trick. This is why I sample the first 40 seconds of the waveform, because in most cases, you do not need to line the waveforms up. Such is the case with the Soundkeeper filesas they both start at exactly the same time.

If you do need to line the waveforms up because you are recording the samples, then you can trim them later in your favorite digital audio editor. It is tedious as it may take a couple of passes before you get it lined up exactly.

Edited to add this section.

I ran another DiffMaker test, this time on Kote Moun Yo? samples from Equinox. I really enjoyed this recording as it definitely has ultrasonic information recorded (i.e. percussion instruments) and is crystal clear sound with very low noise floor. I would say state of the art recording. Great job Barry!

I followed the same process as above. Again, the point in this is to either confirm or deny Monty’s claim that 16/44 is already better than our ears can hear and our sound system can reproduce. 24/192 should contain much more audio information than 16/44, so by comparing 16/44 to 24/192 using DiffMaker will show exactly how much difference there is between the two. In order for me to digitally compare the 16/44 to 24/192, I up-sampled the 16/44 to 24/192. If the R8 Brain resampler I used is doing its job proper, there should be no waveform changes as there is no information being added (or lost!), simply a (lossless) file format change.

Here is what Audio DiffMaker reports as being the difference.

-100dB difference file. It is very similar to my first test above, showing I can repeat the results, even on a completely different song/master.

Here is what the Difference waveform looks like.

And frequency analysis.

As you can see, the frequency plot shows ultrasonic energy, even though it is very low in overall level. Again, I have attached the difference file so you can listen to it. I cannot hear the ultrasonic information.

Part 2 Listening Tests

Given that the difference between 16/44 versus 24/192 is ultrasonic energy, it is important to verify that the gear used can actually reproduce ultrasonic energy. I used my Lynx L22 pro sound card that has a ruler flat frequency response out to at least 50KHz:

I used my Sennheiser headphones with a custom Class A headphone amp that I built from the Audio Amateur from years gone by:

On the right is a toroid transformer feeding a regulated power supply and then my perf boards of the amp itself on the far left. I have measured the frequency response out to +200HKz. The headphone amp has enough clean power that you can place the headphones on the floor opened and crank it up like it was a boom box.

Next step is to verify that my gear can play ultrasonic information properly. These intermodulation test files provided by Monty’s article should be played first on your system to ensure you hear nothing at all. If you do hear tones, pops or clicks, that means the system under test is producing intermodulation distortion.

With my particular computer system, Lynx L22 and Class A headphone amp, I did not hear any tones, clicks or pops. Ok onto step 2.

ABX testing. For listening tests that provide any level of statistical probability, double blind is the only way to go. I used Foobar2000 and the ABX plugin I made sure that I clicked on the Hide Results checkbox before I started the tests.

First up, 16/44 vs 24/192.

Here was the problem with this test. I could just tell by a very small delay when my DAC was switching from 16/44 to 24/192. So I was able to “game” the test:

So I resampled the 16/44 to 24/192 so I could not hear the DAC switch sample rates.

Here are the results:

Obviously I cannot hear the difference. This correlates with the DiffMaker results as well. The difference is so small that I was guessing, even though I was trying not to.

Since I cannot (significantly) measure or hear the difference between 16/44 and 24/192, I tried one more experiment where there is a known difference – MP3.

I took the 16/44 and converted it using the best MP3 codec (LAME) and encoded at 192Kbps bit rate. I used this bit rate as I listen to a lot of music on Zune and this is the default bit-rate when I download the music onto my disk for playing. As you may imagine, there is a reason that Microsoft chose this bit-rate and I will show why shortly.

Now comparing the 16/44 to the MP3 version produces the following Difference file in Audio DiffMaker:

And if I open up the waveform in Audacity:

Frequency Analysis:

I have included the Difference file again so you can hear the results. And it correlates very well with the other two other MP3 difference tests I performed here:

So the $64 million dollar question is, can I hear the difference in an ABX test for 16/44 and MP3?

While I did better than the 16/44 vs 24/192, it is in the territory of guessing :-) Listening closely, I thought I could hear a loss of transients on the percussion, but just barely perceptible to my ears.

Another way I can listen is to use Audio Diffmaker where I can reconstruct the comparison track by adding the difference back to the reference. By incrementally increasing the difference track level, I can easily hear the difference when the difference track is boosted by about +6dB.

I would hazard a guess this is the reason why Microsoft (and others) choose 192Kbps with MP3 as it gives the best fidelity versus file size. And likely the reason why most people don’t complain about it as most people (including me) cannot hear a quality difference, even under ABX testing conditions.


Well, for me, my ears, on my equipment, my test and listening results confirms Monty’s article that 16/44 is enough for my ears. This is also qualified by the science and engineering in the Digital Audio field:

In fact, it may be that even high bitrate MP3’s is enough resolution, but that’s another debate.

Full disclosure, I am 53 years old and given the hearing loss versus age in the chart below, I may not be the best candidate for trying to hear ultrasonic audio information :-)

A quick hearing test from: confirms that I can hear to at least 12KHz, but down at 16KHz. It is no suprise to me why I don't hear ultrasonic audio information:

My perspective is this. If I was going to pick one cause to get behind in the world of music, it would not be over high resolution file formats. It would be the Loudness War.

Almost 30 years ago, the pop band, The Police, created a very popular album called Synchronicity: With an overall Dynamic Range of 15 and the final cut on the album, Murder By Numbers, with a DR of 18 is an excellent example of taking the full advantage of the Red Book standard. The disc sounds fantastic. What happened since then?

The Loudness War in less than 2 minutes:

Given that this is CA, I would think everyone could correlate what they see in the visual representation of the waveform and what they hear. As I have discussed before, there is a direct correlation to what is measured with what is heard – it’s fundamental to the princples of audio. You can see and hear the difference, even over YouTube!

Final thoughts: All of the software used to perform both measurements and listening tests is free. Therefore, if you are curious and want to verify or deny Monty’s (and as it turns out, me too) claim, you can perform the same tests yourself.

Happy listening!
Attached Thumbnails Attached Files

Updated 08-12-2014 at 11:33 AM by mitchco

Personal Blogs


  1. Paul R's Avatar
    May I suggest you use two of the files from Barry Diamant's site instead? Or at the very least, resample 192k material down to 44.1? Since upsampling doesn't restore any information that may have been lost, I think that would be much more interesting, whatever the results.

    Appreciate what you do though. Great stuff. :)

  2. mitchco's Avatar
    re: "May I suggest you use two of the files from Barry Diamant's site instead?"

    That's where the two files came from... click on the link I provided in the post.

    re: "...Since upsampling doesn't restore any information that may have been lost"

    That's the point of the test! The point is to either agree or disagree with Monty's post that states 24/192 is a waste of time because 16/44 is good enough and we can't hear the difference.

    According to my DiffMaker test and listening to the difference file attached, it is indicative that he may very well be right...
  3. audiventory's Avatar
    I make comparing for files 16/44 and 24/192 from

    This files, as I understand, is parallel recorded (no make SRC).

    Playing back from ABX foobar2000 through ASIO-driver -> HDMI -> Receiver (with indication sample rate value for control of accuracy playing).

    Result comparing is 77% correct detection of difference (30 tests).

    It's significantly more 50% (random result for no real difference).


    1) This result only for tested configuration. For other configuration in may different. Depend by signal processing (digital and analog) algorithms.

    2) Result may be better possible if make pause for ear rest.


    For ideal Digital/Analog-equipment (working with exact by Nyquist–Shannon sampling theorem) 16/44 enough for playback. Great oversampling (ideal digital Low Frequencies Filter) must made before DA-conversion only (demand high calculation power - more 41000 FIR-coefficients).

    But at the modern level of development of technics use the high sampling rates easier.
  4. bstcyr's Avatar
    I have always thought that many of my regular red book CDs sound fantastic, and some, not so much. I'm sure all of you have some really good sounding redbook CD's. So it seems that it is possible to make great sounding music recorded at 48/16 and that the resolution is not the problem. The problem then with focusing on raising the resolution is that we're not really fixing the problem, the problem must be something else. Proper recording process, care of microphone set up, lack of compression and many other things that the guys are doing to produce great CD's. Lets push the recording industry to actually get it right and give us the great sounding music that 48/16 is capable of rather than up the resolution and the price but not necesarily the fidelity.
  5. mitchco's Avatar
    Perfect! Thanks for your feedback! This is exactly the kind of data I am looking for.

    audiventory, can you describe what you are hearing that is the difference?

    bstcyr - agree plus no more MP3's.

    Thanks again.
  6. Julf's Avatar
    Hi! Interesting results. Just wanted to check...

    "[i]Result comparing is 77% correct detection of difference (30 tests).[/i]"

    So was that 30 persons doing 1 test each, or did the same person do multiple tests?
  7. audiventory's Avatar
    > can you describe what you are hearing that is the difference?

    One track (192 kHz) has more hi-frequencies than track two (44 kHz).

    It sense as "transparency" and "air". Difference is not great.

    As hypothesis, explain it as more accurate reverse interpolation (digital-analog) at range 3-20 kHz.

    I am not sure what we has clear audio sources (record - processing - mixing) for this experiment.

    A) We make this workflow (record - processing - mixing) in 192 kHz and after make downsampling to 44 kHz.

    B) We make same workflow in 44 kHz.

    A) will sounded better than B).
  8. audiventory's Avatar
    1 person make 30 tests.

    Different 5-10 sec fragments of pair tracks.
  9. Paul R's Avatar
    I may have misread, but didn't you say you upsampled the 16/44.1k file?

    If so, I was suggesting that you use the 192k files as the source. I think that would be more valid than upsampling.


  10. Talk2Me's Avatar
    Bingo! I have no issue when the basics are well taken care in the recording process, and then having the option of listening in hi-res. But, to think one can cut corners it the recording process, and make up for it by increasing the resolution, is just plain wrong. People who complain the 16/44.1 cd medium sounds horrible, are listening to

    1) a bad recording

    2) bad gear
  11. bdiament's Avatar
    An interesting comparison of the data, even if it isn't of the two sources as they are distributed by Soundkeeper.

    I suggest a different test that might more closely reflect how most folks would experience the two: listening.

    The data comparison, showing numbers that would seem to suggest the audible differences are way down in level, does not at all, in my experience, reflect how the files sound when they are played back on a fine audio system (one capable of revealing what 24/192 can do).

    This reminds me of data comparisons between the original CD master and a not-so-good pressing created from that CD master: I can show that when the data are extracted from the disc, they are exactly the same as the CD master. But anyone with a lot of experience creating CD masters knows that what comes back from the plant, when played via a CD player or transport, does not sound exactly the same. (This has been a discussion among my mastering colleagues for some years now and in fact, some replication facilities do a much better job than others - even though data extracted from their discs and from inferior sounding discs will always null, 100%, all the way down.)

    Some folks will not hear any differences between the 16/44 and the 24/192 versions. Experience tells me that different folks have different sensitivities to different aspects of sound, so this isn't surprising. Another factor is the hardware (e.g. in DACs, clocking accuracy and analog stage performance at wide bandwidth) as well as the overall system/room setup will vary. This is one reason why we offer CD versions and don't abandon the medium altogether. I wouldn't want folks spending their money on things they don't hear and have often suggested the regular CD version to some.

    On the other hand, when folks do hear the differences, they can be quite profound. In my case, the 24/192 sounds like my mic feed (something I've talked about elsewhere); the 16/44 doesn't - not in the least. I consider that difference more than a little bit important and quite a bit more significant than the numbers in the null test seem capable of revealing.

    Just my perspective of course.

    Best regards,

  12. mitchco's Avatar
    Re: An interesting comparison of the data, even if it isn't of the two sources as they are distributed by Soundkeeper.

    I downloaded the files from Are you saying those are the wrong files to be using? If so, can you provide me with a location that has the right files? Thanks.

    Re: I suggest a different test that might more closely reflect how most folks would experience the two: listening.

    As a recording/mixing engineer, I have access to multi-million dollar studio facilities to listen in here on the West Coast. I also have a calibrated listening environment in my home that I have measured and documented in other parts of my blog. Listening is not the issue at this time. Please indulge me.

    Our ears are wonderfully adaptable, in fact we are susceptible to auditory illusions As a recording/mixing engineer, I count on the ability to fool peoples ears in believing a recording is in a much bigger space than it really was recorded in for example. Aside from the multitude of digital processors, I can convolve instruments with the impulse responses from famous halls: to make it sound like it was recorded in that hall. Even people that know a particular concert hall intimately can be fooled.

    As described at the front of my post, Monty’s article suggests that 16/44 is good enough and 24/192 is not required. So the purpose of my “experiment” is to correlate what we hear with what we measure and vice versa. If we are hearing a difference, then the audio signal must have been altered in some way. If it has been altered, then the difference can be measured.

    I am looking for both objective and subjective data points to support or deny the claim. Not just objective or subjective data, but both. It has been my experience that there is a direct correlation between the two. That is what I am hoping the outcome of my experiment will be. I have no vested interested one way or another as to the final outcome. It's just an experiment :-)

    I intend to follow-up with the listening tests, but right now I am making some measurements first.

    Barry, if you have preferred files for me to conduct both subjective listening and objective measurement tests on, please let me know where I can download them.


    Updated 05-09-2012 at 03:35 AM by mitchco
  13. bdiament's Avatar
    Hi Mitch,

    The files on the site are the files. Their purpose is for folks to be able to listen to them.

    Your null test involved, per your description, altering one of the files, to wit, upsampling the 16/44 prior to your data comparison. So, I see a few issues:

    1. The files in the comparison are not exactly the files offered on our Format Comparison page. But from the larger view, that's neither here nor there.

    2. The larger issue, as I see it, is performing a null test tells you about the data not about how the files sound when played back and listened to.

    I am all for seeking a correlation between what is heard and what is measured. Where we may differ is in believing the measurements (in this case, a single, one-dimensional measurement) will explain all there is to experience in the listening. I don't think so.

    As I see it, it is like looking at a map of Cleveland and assuming the experience of the map is the same as standing downtown in the city itself. As Korszybski put it "The map is not the territory."

    In my earlier post, I explained how while even though a perfect null (not what you describe but a 100%, all the way down, to the sample null) indicates identical data, this does not in any way mean the two files sound identical. It merely means the extracted data is identical. To assume more is to assume we listen to data. (The only time I've done that is in the presence of a fax machine. ;-})

    If the day arrives when measurements accurately and completely explain everything that can be heard, I would think it safe to select the components of a system by their measurements alone. But to my knowledge, that day hasn't arrived yet. (Would you purchase a system by measurements alone? If so, we must agree to disagree. If not, we have this in common.)

    I applaud the desire for correlation but think we must keep in mind these is still a long way to go in this regard and at best, all we have today are small fragments of what we'd ultimately like to achieve.

    By the way, due to a spike in traffic to our Format Comparison page over the past few days (I believe, due in part to a new review of "Confluence" on TNT-Audio and perhaps moreso to an article on Audiostream), our ISP suspended the site last night. It is now back up but the samples on the Format Comparison page (and a few others) had to be temporarily deleted. I'll work out an alternate means of sharing these post haste - perhaps by tonight, if I can.

    Best regards,

  14. mitchco's Avatar
    Hi Barry,

    Thanks for indulging me. I don't disagree with anything you say. It is just an experiment.

    As to analogy, here is another. Some would argue this is the state of the art in movie production: Watching the behind the scenes footage on how the movie was made is very insightful. No longer can most people tell what was physically shot in the studio or at a site location versus what is digitally enhanced/created by ever increasing powerful computers and sophisticated software:

    My experiment will include listening tests much like audiventory performed below. Both objective and subjective data will get equal weighting in the experiment.

    Re: Confluence. Congrats!


    Updated 05-09-2012 at 03:35 AM by mitchco
  15. audiventory's Avatar
    Hi, Barry.

    Can you show details of producing workflow of 44/16 and 192/24 samples at your Format Comparison page?

    recording (ADC) 192/24 -> processing/mixing (better 192/24) -> downsampling to 44/16? Or parallel processing? Else?

    For me it's not clear, how much these files really approach for comparison of formats? Not including that the difference between 44 and 192 kHz depend both reproducing equipment and listening room.

    To me it is not clear yet: whether it is possible make exact comparison 44 enough or it is necessary 192 kHz.

    If we have correct source audio files with different resolutions, we will compare, all the same, as played back these files at different equipment.

    Best regards,

    Yuri (
  16. bdiament's Avatar
    Hi Yuri,

    The samples on the Soundkeeper "Format Comparison" page represent the best I know how to do for each target format.

    When the first release, "Lift" was done, I was recording at 24/96. For all recordings since then, I've recorded at 24/192.

    Even when the target is a 16/44 CD, my experience has been that I'll get a much better CD ("better" here meaning more like my microphone feed) if the original recording and mastering stages are done at higher resolution. Put another way: Recording at 16/44 will, in my experience, guarantee an inferior CD. More often than not, low level data will get truncated during the mastering process. (Optimal level for digital recording and optimal level for the final product are not the same. Even the tiniest level adjustment will lengthen the digital "word". This simply won't fit into a 16-bit bucket. Even with software that processes internally at longer word lengths, the audible results are not the same as having a high sample rate, longer word length source.)

    So there is no "parallel processing" as anything less than high resolution for the earlier stages of CD production will not give me the best possible CD.

    There is no mixing (or mixing console) on any Soundkeeper Recording. They are recorded direct to stereo using only two microphones in a stereo array. There are no overdubs. Mastering consists primarily of sequencing the tracks, adjusting the space between them and making gain adjustments for final level (because I record with lots of headroom).

    Once the mastering has been complete for the original recorded resolution (again, 24/96 for "Lift" and 24/192 for everything since), the lower resolution versions are created. With a few dozen different SRC and dither/noise shaping algorithms in the toolbox --I'm always testing them--I've found those from iZotope, created by Alexey Lukin, to create results that sound to me (by far) the most like the unprocessed high res original. So their 64-bit SRC is applied to create the 24/96 version and is used for the penultimate step in mastering the 16/44 version. Lastly, their MBIT+ dither is applied to create the final 16/44 version.

    Using a "parallel processed" file for the 16/44 comparison (i.e. something originally recorded at 16/44) would not reflect the real world finished CD -- unless one didn't mind the truncation of low level information on the CD. I find this not only quite audible but quite objectionable.

    As I mentioned above, even if the only target was a 16/44 CD, I would record and master at 24/192 because to my ears, this will create a CD that much better represents the original signals coming from the microphone array.

    So the files on our "Format Comparison" page represent the best I know how to do for each target format. They represent, not a theoretical comparison of recordings made at each resolution but the real world of what is contained in each of the formats available from Soundkeeper.

    (That theoretical comparison is interesting - been there, done that, many times with many different hardware/software combinations - and it is also educational. This is particularly true when one takes the different recordings and takes them through all the steps necessary to create a finished master in each format.)

    I want to reiterate that our "Format Comparison" page is not for the purposes of theoretical comparisons of recordings made at different sample rates and word lengths. (I would think that misleading as it is not, for reasons I stated above, how we make our records and does not represent what our customers are purchasing.) Its purpose is to allow listeners to compare the real thing; the finished products that in each case, represent the best I know how to do in each format.

    Best regards,

  17. audiventory's Avatar
    Hi, Barry.

    Thanks for detailed description of your recording technology.

    Why you make 2 stage SRC for 192/24 to 44/16:

    1) 192/24 - > 96/24

    2) 96/24 + dither -> 44/16

    but not directly 192/24 + dither -> 44/16?


    192 and 44 comparison no make sense practically. Customer-listener hears how his equipment playing back both formats. Thus we can compare both formats in connective: Source(CD/DVD) - Equipment - Room - Ears.

    Theoretical comparison possible for:

    1) Analog precision mixing 1 instrument or band.

    2) Parallel recording in separate files with different resolution.

    3) Listen this files in anechoic room for different equipment.

    4) Process statistic data.

    But we could not remove error playing back equipment for different output sampling rates / bit-depths.

    Best regards,

    Yuri (
  18. bdiament's Avatar
    Hi Yuri,

    I'm sorry if I wasn't clear. Please let me try again:

    In creating our 24/96 and CD versions, we start with our 24/192 finished master. For the 24/96 version, the sample rate is converted from 192k to 96k. The CD version too, starts from the 24/192 version. First, the sample rate is converted to 44.1k, then dither is applied and the word length shortened to 16-bits.

    So, the CD version does not see any 24/96 intermediate step, as you may have concluded from my earlier post. I hope this is more clear now.

    As to the practicality of the comparison, we may have to agree to disagree. If a given recording is available in different resolutions (in our case, all created at the same mastering session), I find it quite practical to be able to compare samples at the different resolutions in order to hear what the sonic differences are between the different versions.

    Further, as I say on the "Format Comparison" page, this is something we believe other music lovers and audiophiles would find of value as well, in view of the fact that most high res versions of releases on the market are more often than not created at different mastering sessions, in different mastering rooms by different mastering engineers (often with different source tapes). In such cases, the listener is actually comparing masterings and not the format differences themselves.

    And nowadays, with all the fake high res being reported from a number of sources, it is nice to allow those who are so inclined to check out samples from our releases so they can see (and hear) the resolution for themselves.

    Best regards,

  19. Julf's Avatar

    "[i]Further, as I say on the "Format Comparison" page, this is something we believe other music lovers and audiophiles would find of value as well, in view of the fact that most high res versions of releases on the market are more often than not created at different mastering sessions, in different mastering rooms by different mastering engineers (often with different source tapes). In such cases, the listener is actually comparing masterings and not the format differences themselves.[/i]"

    Indeed. This is much appreciated! Thanks for making the material available in a comparable form!
  20. bdiament's Avatar
    You're welcome Julf.

    I see from the amount of hits the page and its links get that this would appear to be of interest to many - as I'd hoped it would be.

    Best regards,

  21. audiventory's Avatar
    Thank you, Barry.

    Now all is clearly for me.

    Best regards,

    Yuri (
  22. mitchco's Avatar
    The Scientists and Engineer’s Guide to Digital Signal Processing. Free online version:

    You can read the reviews on

    I highly recommend this chapter (and overall book for that matter) that provides an intro level science and engineering textbook on how ADC and DAC works:

    Topics are, Nyquist sampling theorem , quantization, dithering, aliasing, impulse train, sinc function, antialias filters, single Bit ADC and DAC, delta modulation, etc.

    Includes a mythbuster fact about analog versus digital signals. Plus a few other audio myths are dispelled along the way as it is clear what does and does not affect the audio signal during ADC and DAC.

    Other chapters of interest:

    Sound Quality vs Data Rate:

    “16/44 satisfies even the most picky audiophile. Better than human hearing.”

    High Fidelity Audio:

    “Audiophiles demand the utmost sound quality, and all other factors are treated as secondary. If you had to describe the mindset in one word, it would be: overkill. Rather than just matching the abilities of the human ear, these systems are designed to exceed the limits of hearing. It's the only way to be sure that the reproduced music is pristine. Digital audio was brought to the world by the compact laser disc, or CD. This was a revolution in music; the sound quality of the CD system far exceeds older systems, such as records and tapes. DSP has been at the forefront of this technology.”

    Human Hearing:


    My opinion? I am still gathering data for my experiment plus I need to perform the ABX test as outlined nicely by audiventory and thanks to Barry for supplying state of the art recordings.
    Updated 05-09-2012 at 03:36 AM by mitchco
  23. crisnee's Avatar
    Thanks mitchco for pointing these out. I think I ran across this page/book once, it looks vaguely familiar. I guess I forgot to save it and then forgot about it.

  24. Julf's Avatar
    I haven't read all of Smith's book myself, but from the parts I have seen it seems very good.

    "[i]My opinion? I am still gathering data for my experiment plus I need to perform the ABX test as outlined nicely by audiventory[/i]"

    You might want to wait until next week for the results of my little "just for fun" listening tests. Not for the results, but to take into account all the objections that I am sure will follow, no matter what the result is :)
  25. mitchco's Avatar
    added more info on Audio DiffMaker, another DiffMaker test, and the results of my ABX listening tests.
  26. goldsdad's Avatar
    mitchco, thanks for putting considerable time and effort into the report of your testing.

    I hope you don't mind me pointing out to readers that the DiffMaker "correlated null depth" and the Audacity spectrum graph of a difference track are unrepresentative of the difference at a given instant.

    For example, the 16/44.1 v MP3 has many moments where the difference is between -36 and -24 dBFS, as shown below.

    Click to enlarge

  27. mitchco's Avatar
    no probs at all. In fact, I encourage poking holes at all of this. Just having some fun and sharing the results :-)

    From the DiffMaker help file:

    The Difference signal that Audio DiffMaker makes is the instantaneous difference between the signals.

    When a significant Difference signal is found, there is no guarantee that it might be audible when played as part of the original program material. The program material could mask the difference signal, or the effect may not be of a kind that people can hear. This second case might result, for example, from a device with a slightly nonlinear group delay, or from inserting some extra samples or dropping some from a signal. Audio DiffMaker will highlight differences whether they are audibly significant or not. "Different" doesn't necessarily mean "audibly different"!

    From a technical perspective:

    Sensitivity of Audio DiffMaker to signal changes

    Signal cancellation depth will usually vary with frequency. The sensitivity of the DiffMaker subtractive process to time and relative amplitude errors is easily analyzed mathematically (for any frequency), yielding the following results:

    Phase or Time Sensitivity

    The achievable depth (drop in Difference track energy, relative to the Reference track energy), at any frequency will be limited by the phase error of "theta" degrees existing between the Reference track and the Compared track at that frequency, and will be no better than

    10*log (2-2*cos(theta)) [dB]

    To appreciate this sensitivity to time error, consider an error of just 1/100th of a sample at 48kHz (equal to 208 nanoseconds). At a frequency of 10kHz this is equivalent to 0.75 degrees phase shift error, and it is also the time it takes sound to travel about 0.003 inch (!). From the formula you can infer that if during an acoustical recording a microphone position changes just 0.003 inch that can limit the achievable "depth" at 10kHz to 37dB. In other words, if we wanted to verify that there is no difference between tracks more than 37dB below the existing Reference track levels in a frequency band around 10kHz, microphone to loudspeaker distance should be held to within at least three thousandths of an inch over the duration of the recording of the Reference or Compared tracks.

    Similarly, Audio DiffMaker has to align Reference and Compared tracks to within a small fraction of a sample to possibly be able to cancel to a deep null at high frequencies.

    Difficulty can occur when the sample clock of the digital recording soundcard is not locked to the clock of the signal source. Changes in relative clock speeds that can occur between the two recordings can result in undesirable residual levels. Even quite small amounts of drift can compromise a setup, so it is best to lock sampling clocks together, or provide the Source sound from the same card as is used for recording.

    Amplitude Sensitivity

    The limit to the Difference level depth at any frequency, due to an amplitude error of "G" dB at that frequency from frequency response error or gain error (and neglecting phase shift contributions), is

    20*log(abs(1-10^(G/20))) [dB]

    For example, a volume control shift during a recorded track, resulting from vibration or temperaure drift effects, of just 0.1 dB would limit the Difference signal drop to only 39dB at any frequency.

    Time Drift susceptibility

    Any test in which the signal rate (such as clock speed for a digital source, or tape speed or turntable speed for an analog source) is not constant can result in a large and audible residual level in the Difference track. This is usually heard as a weak version of the Reference track that is present over only a portion of the Difference track, normally dropping into silence midway through the track, then becoming perceptable again toward the end. When severe, it can sound like a "phlanging" effect in the high frequencies over the length of the track. For this reason, it is best to allow DiffMaker to compensate for sample rate drift. The default setting is to allow this compensation, with an accuracy level of "4".

    Gain Drift susceptibility

    Usually a lesser problem, Gain Drift is a varying signal gain during the time the recordings are made. The gain drift may result from mechanically or thermally induced changes in circuit components slowly drifting over time, or from variations in voltage references used by the A/D or D/A converters in the soundcard being used.
    Updated 05-09-2012 at 03:37 AM by mitchco
  28. robocop's Avatar
    I concur with Barry re mastering. Two recent purchases only available on CD 16/44 bear out the process of mastering at 24/192 and converting to 16/44 being superior to recording at 16/44. I've recently also converted two 16/44 downloads to 24/192, although not a large difference never the less they are much easier to listen to at 24/192.

    I'm 57 and have similiar hearing deficiencies to Mitch. Even though we can't clearly hear high frequencies above say 12khz, I've found since adding super tweeters that go up to about 70khz I can't live without them. They add something to the harmonic structure of music that affects all frequencies down to the bass. When I turn them off, its noticeable immediately even by casual listeners. 16/44 is never going to give it all to us.

    Music is a complicated structure of interrelated sounds that need to be recorded from 0 to 100k to give us some semblance of accuracy. Any frequency missing or cut-off will affect the sum of the whole sound.
  29. mitchco's Avatar
    hey robocop, thanks for your comment. What super tweeters are you using? I am interested.

    I recently upgraded my DAC to a Lynx Hilo and was quite surprised at the SQ difference I heard compared to my Lynx L22 (a ten year old design, but still sounded pretty good), especially at 24/192.

    Have a look at this very interesting paper on the Theory Of Upsampled Digital Audio for possible answers to this dilemma.

    Once I have some time, I intend to re-run my tests here with the Hilo and see if the results are the same.

    Cheers, Mitch
  30. timanderson's Avatar
    Hi Mitch, did you repeat your experiment with the Hilo and if so what was the outcome? I agree with the downsample, upsample, compare with original approach. This seems to me fair, as upsampling to 24/192 and playing the upsampled file is merely another way of playing a 16/44 file, you could do the same with any 16/44 source.
  31. mitchco's Avatar
    Quote Originally Posted by timanderson
    Hi Mitch, did you repeat your experiment with the Hilo and if so what was the outcome? I agree with the downsample, upsample, compare with original approach. This seems to me fair, as upsampling to 24/192 and playing the upsampled file is merely another way of playing a 16/44 file, you could do the same with any 16/44 source.
    Happy New Year Tim.

    I have not got around to this yet. I have been sidetracked by several projects, including refinishing my speakers over Xmas and putting them back together today. Probably in a month or so.