數碼取樣試聽分享

昨天上載了前兩天的試聽報告到computeraudiophile網頁，現在將報告放在這裡與大家分享。
Objective:
The objective is to obtain preliminary answers to three questions through comparative listening, namely:
1. Whether participants of the listening session come to the same observation as some audiophiles say that the CD format sounds as good as high resolution format? Or put in another way, whether participants come to the same listening observations as some audiophiles suggest that CD tracks played on a good transport sound as good as or even better than high resolution tracks played by a computer with the same external DAC?
2. Whether CD tracks sound the same as their ripped offsprings that have been upward sampled?
3. Whether PC sounds the same as Mac, with the same tracks being decoded by the same DAC?
Observations:
1. CD tracks (those under test) sounded from worse to very much worse than high res tracks, even though both came from the same master, this being the case even with an excellent transport such as the MSB playing the CD. They sounded considerably worse when played at their native 16/44.1 rate and less worse on many counts when they were upward sampled on the fly by the MSB DAC to 24/352.8.
2. CD tracks (those under test) even upward sampled by MSB sounded worse on some counts than their upward sampled offsprings played by a Mac. The differences heard might perhaps be attributable to differences in the upward sampling algorithms, between the one in the Korg used for tracks in the Mac and the other in the MSB DAC.
3. As regards PC versus Mac, on some counts it was a toss between the two whereas on other counts, Mac sounded better or vice versa. Overall the Mac scored higher insofar as the tracks under test were concerned. On the assumption that the differences heard are genuine, contributing factors may perhaps be hard disk versus SSD, Windows 7 versus Lion OS X, Jplay versus iTunes and upward sampling algorithm, Korg versus MSB.
ADDITIONALLY, it was found during the session that:
4. Somehow the 2L tracks in the CD sampler sounded louder than the same tracks in DXD format copied from the stereo masters of 2L.
5. A couple of tracks played by the PC sounded louder than the Mac. However no further investigation was conducted with the purpose of finding out whether it was the PC that put out a louder sound than the Mac or Jplay raised the level higher than iTunes. If it is Jplay, then its developer is clever in enhancing this feature because quite often listeners come to a conclusion though incorrectly that the louder one sounds better than its softer competitors. It should be mentioned that all tracks in the Mac have sound level set at centre position (displayed under Options) which means the level is intact, there being no tampering after copying from DVD-ROM or downloading from the web.
6. Some tracks showed a big difference when switched between two absolute phases whilst some showed very little such as the two Unipheye string quartet tracks.
Time, place and participants
Day before yesterday, 19th Oct 2011 at Ben's place. His listening room is about 250 to 300 sq ft, rectangular in shape and having one wooden acoustic treatment panel, 3 x 5 ft, placed at each of the two corners of the back wall. Speakers are positioned almost halfway from the back wall, and each being 3 to 4 ft from the side wall.
Participants were only two, Ben and the writer. Originally Doctor John who runs a blog of his own was to come along but could not make it eventually because of an urgent matter.
Setup and methodology
Ben's audio system facilitates staging of such comparisons because:
his MSB Signature IV DAC comprises 3 separate components, a transport, a DAC and a power supply. This DAC has USB input which is asynchronous and also has upward sampling and phase change selectors that may be switched on or off by remote from the listening seat. More details on this MSB transport and DAC may be found in the manufacturer's site
http://www.msbtech.com/products/dac4home.php
Speakers are Volent Paragon VL 4. Details are in
http://www.volent.com.hk/product-pag...ctpage-vl4.htm
A Pass Labs stereo amplifier is used for the midrange and up whilst two non-brand mono amps (also solid state) drive the bass units. Pre-amp is non-brand as well, vacuum tube.
Ben's PC is IBM T60 (windows 7) with Jplay (3.4) and the writer's is MacBook Air (Lion 10.7.2) with iTunes (10.5). The IBM computer is normally connected by way of a high end non-brand USB cable to the DAC whilst the Mac uses a side street one. After some time we started using the Mac's and another of Ben's cheap USB for the PC as well because high frequencies sounded more solid with them. All other cables and interconnects of the audio system remained intact throughout the session, for example, the MSB CD transport was connected to the DAC by I2S purchased as an option from the manufacturer.
As MIDI of the Mac showed it was possible to output at 24/352.8 (DXD) after recognizing the MSB DAC, and indeed the display of the MSB indicated tracks of this format were at play, this choice stayed put throughout the session. The writer was pleased to find non DXD tracks saved in the Mac could be read by the DAC as 24/352.8 when the MIDI was set at this rate. This possibility suggested OS X/iTunes upward sampled them before outputting to the DAC.
One matter that complicated listening comparisons was the sound level differences between competing tracks (Observation 5 above) as well as similar differences between PC/Jplay and Mac/iTunes (Observation 5). These differences necessitated turning the volume knob of the pre-amp up and down every now and then so as to equate roughly the sound levels, quite unscientific but no better way on hand.
The track list for listening under all 3 Parts as well as the suggested features in each of them to be focussed upon were circulated in advance to participants so that they might get familiar with the features at home and be able to pick out differences quickly during the session. As Ben had higher interest in comparing the PC and the Mac (Part 3), it followed that more listening was carried out under this Part than originally intended. That was because he had previously made comparisons between the CD format and high res (Part 1) and in fact he always had the upward sampling function turned on so as to improve the CD sound. The tracks listed below were the actual ones played, after removing some tracks from the original Parts 1, 2 and 3 list and adding new ones to Part 3.
Comparative listening under Part 3 was complicated further by different options available. For example, there are two modes under Jplay, Beach and River, with each introducing either benefits or drawbacks, varying from track to track. Also the upward sampling function in the MSB expands the soundstage and projects the sound forward but at the same time softens the impact and clarity of instruments. This function on the whole provides sonic benefit to the PC/Jplay because of their limitation to 24/192. On the other hand, as the Mac/iTunes support output at 24/352.8, tracks if further processed by this function gain only an expanded soundstage but suffer artificialness in instruments and voices. In view of all these complications, it seems desirable to have a further session some time later. Prior to that event Ben will go through a dozen or more tracks, first identifying which option, Beach or River, that gives a better sound whilst also marking down the good and bad features arising from the MSB upward sampling function. In this way the dedicated session for comparing the PC and Mac will be streamlined and focussed.
Rationale for choosing the tracks and limitations
On the face of specifications the CD format has less extensions in both high and low frequencies than high res, and also has less density in the midrange because of lower sampling rate and word length. The tracks chosen for the session contain high and low frequency extensions and dense midrange with a view to revealing differences, if any, more easily. As resources are limited, variables uncontrolled and time is constrained, the listening results are perforce observations obtained in that particular session, and not definitive conclusions applicable to most other circumstances. The purpose of sharing the observations is to set out in details how they are arrived at. Hitherto audiophiles belonging to different camps only talked about their views without stating exactly how they were formed. Also it appears no reviewer in audio magazines has touched on these issues, nor has any thread in blogs covered them adequately. Well cynics may say these observations are flawed because of placebo effect as listening was not double blind. Others may also question the choice of tracks and the effect of variables on listening that lead to the observations.
Phase (polarity) and Jplay modes
The tube pre-amp inverts the phase and it is unknown whether tracks in the chosen CDs also have phase change during mastering and mixing. Therefore it is prudent to ensure the phase is the same during comparison of CD and computer (Mac) tracks. Hence every play from either of them was to go through a phase test prior to focussed listening. For Part 3 involving a comparison between PC/Jplay and Mac/iTunes, tracks in the former were run in both Beach and River mode so as to obtain a better choice for making the comparisons. Every track under test in this Part also had to go through phase checking.
Listening notes for Part 1, CD versus high res from same master
Note: CD tracks were played in the MSB transport and decoded by the MSB DAC. Tracks from 2L were all in DXD format (24/352.8) and saved as such in the Mac. They came out from the Mac as 24/352.8 and were decoded by the same MSB DAC as the CDs.
*2L 038 Mozart Violin Concerto (CD track in 2L Sampler)
In the 16/44.1 format, the soundstage was squeezed, layers of players were compressed, and the violin and strings sounded shrill and coarse. Switching on the upward sampling function in the MSB to 24/352.8 improved the sound considerably. However, the upward sampled CD sound still could not match that of the native DXD on the counts mentioned and it also had less details. The CD sound was louder and so when playing the competing track in the Mac the volume knob of the pre-amp had to be turned up arbitrarily to match. Whilst listening Ben noted the Mac track had better sound than the same one in his PC, based on what he had previously heard. Thus there followed several comparisons between the two and these are set out in Part 3.
*2L 077 Harmonica/Organ (CD track in 2011 HK AV Show album)
In its native format, the CD track had a smaller soundstage, constricted bass, subdued high frequency extensions and softer impact of the low end. Upward sampling to 24/352.8 with the MSB improved the sound but still no match for the native DXD format in the Mac. Again Ben noted the Mac track sounded better than the one in his PC thus further comparison between the two followed later. To provide time slots for adding comparisons between the PC and Mac under Part 3, two tracks originally scheduled for Part 1 were skipped.
*Ting Ting CD by Bondy Chiu Hok E: 愛如潮水 track (track in the Mac was originally 24/96 and then upward sampled before being saved)
After listening briefly to the CD format, we quickly switched in the upward sampling function so as to improve the sound. Even so the guitar and the saxophone still stuck to the left and right speakers whilst the voice was way back in the centre near the wall. Switching over to the track in the Mac, the guitar and saxophone moved back somewhat from the speakers and the image was more 3 D. The guitar sound was more vivid and the female voice more consolidated (focused and less diffused). Both CD and Mac tracks rendered the voice nicely, with no strain, edginess, shrillness or compression. There were however three shortcomings in this track: a) the guitar, though recorded by close microphone placement, did not have adequate harmonics to suggest without doubt it was an acoustic instrument, lacking similar hints as in other well recorded guitar tracks; b) the singing skill of the female singer was below top rate, for example, when compared to the various artists in the Linn Sampler 5; and c) the saxophone was recorded unnaturally close to the microphone.
*Audiophile Jazz Prologue 3 CD album, Lush Life (track in the Mac was downloaded as 24/192 but no information as to whether this was native or upward sampled; Ben also had the 24/192 DVD-ROM but no information there either)
The CD sound, upward sampled on the fly by the MSB, still had several words sounding like overloaded, having strain, compression and ringing: "washed away", "then you came along", "me to madness", "tinged with the sadness". The track in the Mac also had problems with these underlined words but only slightly noticeable whilst the accompanying guitar was more vivid and the voice more consolidated. The guitar sound still did not have enough harmonics to suggest it was definitively an acoustic one.
Listening notes for Part 2, CD versus its ripped offspring
* 八音和 ((in the 2010 HK AV Show CD)
This track was recorded in a church with 2 microphones into a Korg recorder in 5.244 DSD format. Instruments used were Chinese wooden blocks, cowbell, hanging cymbals, hanging gong, hand held clash cymbals, Chinese drums slightly larger than bongo, bongo, and floor standing large Chinese drum.
The following features were focussed upon: resonances of the two cymbals as well as those of the hand held crash cymbals; sound of sticks rolling on the cymbal bowl from ride to crash; impact of the floor drum.
Ben preferred the side street USB cable to his high end one in playing this track and he considered the CD track when upward sampled on the fly had more convincing highs than the ripped track in the Mac. On the other hand he agreed the Mac track had better drum sound that was more open (less constricted) and solid with impact.
*Jaleo Solea (in the JVC Flamenco CD album)
The CD track being upward sampled by the MSB when compared with the pre upward sampled track in the Mac: the Mac track had a more realistic sound of guitars as well as castanets; a soundstage with guitarists and flamenco dancers placed in layers; less distortion and compression in the loud male voice and tighter and more impact of boots of dancers stumping the wooden floor.
*Let's Call The Whole Thing Off (in the Harry Met Sally CD album, Harry Connick Jr)
CD track played with upward sampling when compared with pre upward sampled track in the Mac: voices were on a par between the two; the Mac track had tighter and louder rendition of the double bass and cleaner and sharper sound of brushes on the snare drum.
*Where Have All The Flowers Gone: solo singing by Pete Seeger with no accompaniment (Let's Folk CD album from Sony)
In regard to the two comparative tracks Ben and the writer had different preferences. Ben considered the CD track more musical (words flow more vividly towards end of phrases) but he agreed with the writer that the male voice was more consolidated with a tighter focus. As regards preference, he valued the musical aspect more because he considered the consolidation and focus of the voice in the Mac track inadequate to tip the balance.
*Testament of Freedom (in the Reference Recording HDCD Sampler)
The CD track being upward sampled by the MSB had a larger soundstage but image was more laid back. The Mac track provided a clearer head count of the chorus members; brasses were less compressed and blossoming forward; the bass drum was more solid with impact.
Listening notes for Part 3, PC versus Mac
*2L 038 Mozart Violin Concerto
As mentioned in Part 1, Ben considered the Mac producing a better sound of this track than his PC even though both were of DXD format. A series of back and forth comparisons were carried out. First was the swapping of the USB cord between the PC and the Mac. He still preferred the sound from the Mac. Then the track in the PC was copied to the Mac and played there under Desktop for comparison with the original one in the Mac. Ben still preferred the Mac track. So it was then copied to the PC and played there. Ben still preferred the Mac track thus he came to the view that 2L had done some processing to the track prior to uploading it to the website for free downloading. The writer always had the view that downloading from the web and copying from DVD-ROM entailed differences because the former had to undergo chopping down, compression and reconstruction during the transmission and delivery processes. Also his PC/Jplay was restricted to 24/192 and the track depended on upward sampling to go back to 24/352.8. On the other hand the Mac track was played and decoded at its native DXD rate thus it sounded better.
*2L 077 Harmonica/Organ
The comparison between PC and Mac was slotted in because of Ben's observation as mentioned in Part 1.
After back and forth listening between the two, Ben considered the Mac track sounding much better in terms of a more projected sound that filled up a larger area of the listening room and also the tighter and stronger impact of the bass notes. However he noted in the PC rendition movements of the player's mouth on the harmonica and this nuance was absent in the Mac. But the writer was unable to catch this virtue in both tracks even after Ben mentioned it. Ben knew this track well, it being one of his favorites and he often played it to friends visiting him.
*Rachmaninoff Symphonic Dances 3rd movement, HRx 95
Jplay did not have fast forward and listening was stopped after several minutes so that the Mac might take over. Under the River mode, strings were shrill and coarse. The Beach mode improved the sound considerably. It was a win some lose some situation between the PC and the Mac.
*Mendelssohn String Quartet 2nd and third movements (Unipheye)
Ben preferred the sound of PC track because the first violin sounded louder thus giving a better anchor for assessing the image positions of the four players. On the other hand the writer considered the violin sound too shrill and bright. Ben agreed with the writer that the Mac gave a more realistic sound of the cello, both plucking and bowing. Ben also favored the attack and impact of the strings at the beginning of each phrase as rendered by the PC. The writer was unable to capture this virtue during listening.
*2L 049 Beethoven Piano Sonata
Ben preferred the sound of the Mac track to the one in his PC that he downloaded from the 2L web. In the Mac track he noted more bass resonance reflected from the wooden bottom of the piano body. As this track had a lot of banging that was good for demonstration but not for regular listening, the writer played him another track from the same album. Ben agreed that in this second track the pianist played the piece with more feeling and devotion that he termed more musical. He could note the piano keys when struck by the pianist produced a sound that was crisp, clear and solid with impact. Further the piano had a more 3D image projected into the listening area, representing a better rendition of the reflection from the sound board. Ben also noted resonance of the strings when struck by the hammer and the rich harmonics arising from the resonating strings. Ben said he could visualize the pianist playing cross-handed in some passages (right hand playing keys on the left line and vice versa). But such perception was beyond the writer's listening capability.
xxxxxxxxxxxxxxxxx