A few words of description regarding the echo/ripples injected in to the Echo experiment test files. As the experiment introduction states, the files are in sets of triplets, each possessing one of three levels of injected echo/frequency-response ripples. The objective magnitude of these is very low, just as in commercial OS FIR interpolation-filter chips. We are attempting to realistically assess their detectability in typical OS DAC application. Here is Marcel's basic description to me of the file preparations.
"I propose to make three versions of each track, one that's just dithered and requantized, one with SAA7220-like echoes and dither and requantization and one with echoes equivalent to +/- 0.1 dB ripple, dither and requantization."
Marcel, feel free to correct my factual errors, or to add further comment to this post.
"I propose to make three versions of each track, one that's just dithered and requantized, one with SAA7220-like echoes and dither and requantization and one with echoes equivalent to +/- 0.1 dB ripple, dither and requantization."
Marcel, feel free to correct my factual errors, or to add further comment to this post.
I just upgraded Amanero module on my Audio GD R2R-11 with Crystek CCHD-575 oscilators, noticed immediate improvement. Not a jaw dropping, but noticeable. Reading all posts from the beginning while sound is still improving. In meantime I had to resolve strange occasional quality problem. I suspected my hearing not right, but it turned out, it was a failing led light spreading a noise. When it eventually died, I noticed that a lower wattage replacement is many times brighter. Now SQ is stable and I started testing the last round...
Well, on some tracks I hear minimal differences, but I don't know what to focus on, it makes no easy, pleasure generally is the same. I put some tracks in a loop while doing something else, all beatiful music I do appreciate. Listening for few hours and waiting for sympthoms of fatigue. Nothing... Swtching to the infamous Topping D30 and after 10 minutes I am losing ability to recognise details. Everything becomes harsh and unpleasant after a short time.
I think we don't test right things. I don't have good hearing and I am close to 70's, but I am traditionally very sensitive to Delta-Sigma distortions. This sound is losing all reverbations on a decay, produce false harmonics. As opposed to these files, Delta-Sigma sound is completely different, therefore my question is, which upsampling 88.2k options were used in the four files test? Noise shaping is the most suspicious in my view.
Well, on some tracks I hear minimal differences, but I don't know what to focus on, it makes no easy, pleasure generally is the same. I put some tracks in a loop while doing something else, all beatiful music I do appreciate. Listening for few hours and waiting for sympthoms of fatigue. Nothing... Swtching to the infamous Topping D30 and after 10 minutes I am losing ability to recognise details. Everything becomes harsh and unpleasant after a short time.
I think we don't test right things. I don't have good hearing and I am close to 70's, but I am traditionally very sensitive to Delta-Sigma distortions. This sound is losing all reverbations on a decay, produce false harmonics. As opposed to these files, Delta-Sigma sound is completely different, therefore my question is, which upsampling 88.2k options were used in the four files test? Noise shaping is the most suspicious in my view.
Last edited:
I'm uncertain whether your post is referring to your experience performing our Echo experiment. If it does, then my first question is, are you using a NOS, or a OS DAC?
As far as what to listen for in the test, and aside from listening for most pleasure, or the least displeasure, you could listen for the sound which characterizes OS playback versus NOS playback, to you. The experiment presupposes that you subjectively hear a characteristic difference between NOS and OS. If you don't normally hear any difference, then you won't hear a difference among the test files, which are processed to simulate an particular artifact of Equiripple type OS interpolation-filters. In such case, you should not submit a test report.
If, instead, you typically do hear a difference in NOS and OS playback, then perform the Echo-test and submit a report with the files ranked as you feel that hear them. Don't agonize over the ranking.
The Alexander Gibson - Witches Brew files are difficult to determine in terms of one being better than another. In researching this, it seems the original was recorded in 1958 or so, hence I am wondering about the quality of the initial conversion to digital. Do you know were this came from?
R2R-11 is NOS, a pure NOS, no preprocessing, as there is no appropriate hardware. Just Amanero USB receiver and a simple CPLD driving ladders, there is no sophisticated analog filter as in early Sony players.
I am having a problem identifying what you mean by saying ECHO. A short echo like in a room acoustic, may be a longer one. I hear some difference on two of four samples, but I think, I made it clear, pleasure is the same. There is something else affecting Delta-Sigma sound. A difference is on Keith Jarrett track (I have this recording in 24/96k format) and Scarlatti. On this point I can vote for the best, not sure I will be able to pickup the worse item. John Williams is overprocessed in my opinion, perhaps original was made for multichanel and then downsampled to stereo. It can mask any further filters applied. Saint-Saens needs further listening, I realy love this recording, I expect a clue should come from this track.
Before a clock upgrade I made number of resampling tests with Foobar/SoX and finally decided to play in a native format, resampling was not beneficial for me. Did it sound different? Yes - but minimal and not better. I also own a Nobsound 8xTDA1387 DAC and it sounds like NOS, while is probably oversampling 4x.
Last edited:
Actually if I could hear anything (still not sure) I have the same feeling about the tracks. I maybe heard somewhat clearer differences in the Keith Jarrett and Scarlatti tracks. It all felt like guesswork TBH, but at least if felt more like hearing-aided-guesswork for those two 
. But we will see once Marcel lifts the curtain...
. But we will see once Marcel lifts the curtain...
Performed by; New Symphony Orchestra of London, originally recorded in 1958.
Now this, from the liner sheet: "Transferred from the original stereo source tapes. The transfer utilized a 20-bit analog to digital converter. The most advanced reclocking is used to ensure the absolute lowest level of jitter. We employ a minimum number of stages, and as little processing as possible."
Let me suggest what may be a less agonizing and hair-splitting means for determining your level of listening fatigue. Which, for me, is the greatest, and most significant, characteristic difference between typical OS, and NOS.
A large difficulty in judging, so called, listening fatigue is that it seems to be largely psychological in nature. It's not like judging frequency balance, or resolution of detail. It also may take a long time before you begin to perceive an outright feeling of being tired of listening. In my experience, what usually occurs first, is a feeling of boredom or disinterest with the playback. Typically, manifesting as an difficult to describe desire to switch to some other track, or to stop listening and do something else.
On the other hand, fatigue-free playback encourages listening to a given track for a longer period of time. Even should the music genre not be what you would typically listen to, you don't feel bored, or the desire to stop playback, and do something else.
The experiment is not to test anything so obvious and gross as discretely audible echoes, like someone yelling in the mountains. This is a test of the often subtle subjective difference between OS and NOS. In other words, if you cannot hear a difference between OS and NOS in regular listening, you very probably will not hear one in the experiment.
What I've noticed so far regarding the echo experiment's triplet files, is that there always seems to be one file of the three which encourages me to listen to it for a longer period than the other two. Likewise, there always seems to be one file of the three which compels me to listen to it for a shorter period of time than the other two.
A large difficulty in judging, so called, listening fatigue is that it seems to be largely psychological in nature. It's not like judging frequency balance, or resolution of detail. It also may take a long time before you begin to perceive an outright feeling of being tired of listening. In my experience, what usually occurs first, is a feeling of boredom or disinterest with the playback. Typically, manifesting as an difficult to describe desire to switch to some other track, or to stop listening and do something else.
On the other hand, fatigue-free playback encourages listening to a given track for a longer period of time. Even should the music genre not be what you would typically listen to, you don't feel bored, or the desire to stop playback, and do something else.
The experiment is not to test anything so obvious and gross as discretely audible echoes, like someone yelling in the mountains. This is a test of the often subtle subjective difference between OS and NOS. In other words, if you cannot hear a difference between OS and NOS in regular listening, you very probably will not hear one in the experiment.
What I've noticed so far regarding the echo experiment's triplet files, is that there always seems to be one file of the three which encourages me to listen to it for a longer period than the other two. Likewise, there always seems to be one file of the three which compels me to listen to it for a shorter period of time than the other two.
Last edited:
With it's massed brass, I've always found this CD very difficult to get non-shrill sound from via OS, and from this track in particular. Via NOS is the closest I've come to it sounding natural. This is also the reason I've included the track in our test.
Now some details about the CD: Title; 'Raiders of the Lost Ark'. The London Symphony Orchestra, John Williams conducting. Mastered for CD by, Steve Hoffman. I did not spot any liner notes, however, on recording technique, or equipment.
Last edited:
I submitted my results to Marcel. All the files where generally better than in previous tests, with the "Raiders of the Lost Ark" and the "Scarlatti" tracks being more difficult to guess at. The massed brass was hard to take as you indicated, though a friend commented this is often the case in real life.
Gerrit
The provenance is helpful, as I like to compare to the CD/record, or same files on Tidal if possible.
Gerrit
Had been listening in a repeat mode, waiting for a fatigue (remember I am very sensitive to Delta-Sigma distortions) and nothing comes out. Sorry, it doesn't work with my brain.
However maybe I get a clue on the Saint Saens tracks. The first 19 seconds reveals a difference in gong reverbations. When a gong is not hit anymore, it happens at 0:13, focus on a gong reverbations and constant frequency shift. When a string section comes up something happens with a gong...
On the one track frequency of the tone is constantly changing during pulsation. It looks to me the most natural behaviour. Sound is also more pronounced, full body, the strongest reverbations.
On the other track frequency of a tone is becoming less transitional and pulsation becomes more regular.
On the last track frequency shift disappear, there is still pulsation, but it becomes slightly blured with tempo of the music.
However maybe I get a clue on the Saint Saens tracks. The first 19 seconds reveals a difference in gong reverbations. When a gong is not hit anymore, it happens at 0:13, focus on a gong reverbations and constant frequency shift. When a string section comes up something happens with a gong...
On the one track frequency of the tone is constantly changing during pulsation. It looks to me the most natural behaviour. Sound is also more pronounced, full body, the strongest reverbations.
On the other track frequency of a tone is becoming less transitional and pulsation becomes more regular.
On the last track frequency shift disappear, there is still pulsation, but it becomes slightly blured with tempo of the music.
That's alright, sajunky. Go ahead and rank the Saint-Saens files as you prefer them, and rank the other track's files, where you have absolutely no preference, in any order. If you were listening all the way through entire files on repeat mode, then I suspect that you may have put too much effort in this
.I'm curious to know, do you normally hear any characteristic difference between NOS playback and OS playback in casual listening sessions?
Last edited:
For casual listening, I don't believe so.
My point was that, should a person not feel they hear a difference between NOS and OS playback under normal listening circumstances, then they probably wouldn't hear one within the context of a listening test either. Such a difference, is what the test is intended to revel.
On the other hand, a person who does feel they normally hear a characteristic difference, of which there are many such people, probably is a person who primarily owns a NOS DAC.