You quoted a bit by a poster called 'link' but your link pointed to a post by 'zephirus'. I couldn't find a post by 'link' anywhere on that thread.
The statement “sounds better” is ambiguous. I see two interpretations.
The first one is that subjectively there is a preference for music played on LPs versus CDs. There's not much that can be further said about this, except that an individual’s preference is just that. It is their unique preference and no one can say whether that is good or bad, it just is.
The second interpretation is that there is some sense that the music played through LPs is more accurate than the same music played via a CD. This opens up the claim to objective measurements, since accuracy in audio engineering terms is about things like frequency response, distortion, phase errors etc. On that basis what we know about the measurements from the CD transport system is that they are superior to those from the LP. Not only superior in the absolute sense of measured difference through measuring devices, but the difference is such that, from what we know from psychoacoustics, we would expect people to be able to detect a difference between the two formats.
There may be a counter argument that even if there is a difference in accuracy between the LP and CD formats, there is some sense a preference for the harmonic distortion and other artefacts inherent in the LP format - it is these artefacts that people prefer because they "sound better". The problem with this argument is that in the accuracy versus preference tests that researchers like Toole have conducted, there is a general consensus that flatter frequency response and lower distortion is preferable just because it brings the listener closer to the musical event and is preferred because the music sounds more natural through systems that have flat frequency response and low distortion.
In addition, all the discussion about pressing errors in CDs etc misses the point that the audio errors inherent in the LP format are much larger than those that may be present in the CD format.
The first one is that subjectively there is a preference for music played on LPs versus CDs. There's not much that can be further said about this, except that an individual’s preference is just that. It is their unique preference and no one can say whether that is good or bad, it just is.
The second interpretation is that there is some sense that the music played through LPs is more accurate than the same music played via a CD. This opens up the claim to objective measurements, since accuracy in audio engineering terms is about things like frequency response, distortion, phase errors etc. On that basis what we know about the measurements from the CD transport system is that they are superior to those from the LP. Not only superior in the absolute sense of measured difference through measuring devices, but the difference is such that, from what we know from psychoacoustics, we would expect people to be able to detect a difference between the two formats.
There may be a counter argument that even if there is a difference in accuracy between the LP and CD formats, there is some sense a preference for the harmonic distortion and other artefacts inherent in the LP format - it is these artefacts that people prefer because they "sound better". The problem with this argument is that in the accuracy versus preference tests that researchers like Toole have conducted, there is a general consensus that flatter frequency response and lower distortion is preferable just because it brings the listener closer to the musical event and is preferred because the music sounds more natural through systems that have flat frequency response and low distortion.
In addition, all the discussion about pressing errors in CDs etc misses the point that the audio errors inherent in the LP format are much larger than those that may be present in the CD format.
Yet when I rip a CD which is not a Chinese pirate version using dBPoweramp, it checks the data against 'accurate rip' and signals when there are no errors in the track. Its common to get an accurate rip (zero errors) and my rip takes place at up to 48X. So whatever my CD drive's quality (I have several) there are generally no errors when ripping a disk at high speed. As a control that the software isn't cheating, Chinese pirated disks do usually give errors.
You said earlier that your findings were of significant uncorrectable errors so what CD player were you using? It must be poorer quality than the cheapest CD drive in my PC because its giving you significant errors at 1X playback.
In a non scientific DBT theres a proffesor who asks a large class if they prefer a low bitrate MP3 over a CD. The large majority prefer the mp3 for pop/rock music. This is one reason I give little creedance to the mantra " It sounds better "
abraxalito,
Remarkable results indeed!
You must have some of the most exotic gear in the world if you're able to get no errors or significant uncorrectable (C2) errors. Evidently, you're able to get Block Error Rates that surpass anything that the laboratories have achieved so far. As I mentioned on another thread, Puget Sound Studios (a SOTA recording/mastering studio) produced a CD for our speaker contest that had a Block Error Rate of under .5 (point five). JVC's R&D Lab in Japan is sometimes able to get a 1X BLER of under .3!
ISO 10149 requires a frame error rate less than 3%, or a 1X BLER of 220 per second. Most Audiophile labels are getting around 40 to 80 1X BLER, although the mainstream commercial labels are usually somewhere between 100 to
200, which within the Music Industry is considered acceptable.
Do you test for BLER yourself or do you outsource it?
Best Regards,
TerryO
No, just bog-standard IDE and SATA drives in my PC. Nothing fancy at all.
Do please explain what a 'Block Error Rate' is for ignoramuses like me.
I have no idea what you're talking about. Perhaps you didn't read my post carefully enough?😀
Really? How did you determine that from his post? I can't see yet how BLER is even relevant to what I posted, but hopefully he'll be back to explain.😀
I'm not sure if Terry is talking about reading or writing.
dbPoweramp does rerip frames if it finds an error. I believe this is triggered by BLER and other things. The Accurate Rip (part of the program) compares your rip to others first. If it's exactly the same, then it passes the track as good. This does not mean it is error free, just that it exactly like rips by other folks. If it is, you can be pretty sure you are getting exactly what's on the CD, factory installed errors and all.
dbPoweramp does rerip frames if it finds an error. I believe this is triggered by BLER and other things. The Accurate Rip (part of the program) compares your rip to others first. If it's exactly the same, then it passes the track as good. This does not mean it is error free, just that it exactly like rips by other folks. If it is, you can be pretty sure you are getting exactly what's on the CD, factory installed errors and all.
Sounds the same as my understanding. The chance of getting the same rip as someone else when reading errors are present is infinitesmally small. So I conclude that any issues in relation to the reading back of the CD are non-issues - the rip really does accurately reflect what's on the CD. Thus BLER and laser mis-tracking and c2 errors and the like are completely irrelevant. All this on a $20 drive in my PC.
Last edited:
The ISO 10149 standard is for optical discs not the CD audio Red Book standard – so how is the ISO 10149 relevant for CD audio?
Also I didn’t know what BLER was so I found some text from Frequently Asked Questions About Compact Discs:
BLER is an abbreviation of BLock Error Rate, but really represents frame error rate. One frame represents the smallest integral data package, and contains 24 bytes of data along with sync, subcode, Q parity, and P parity. Data is read from a CD at the rate of 7350 frames per second in a 1X player. After alternate bytes are delayed by one frame, BLER measures the rate of bad frames that contain one or more read errors. If one percent of the frames contain errors, then BLER will be 73.5 per second at 1X.
More:
The weakness of BLER is that it counts error frames, not bad bits or bytes.
So, to what extent are these errors corrected by the error correction algorithms? And are errors within the Red Book ISO standard audible?
I've done a bit more reading up on BLER, using this Stereophile article. BLER applies prior to correction - Robert Harley found no E23 errors (errors which can't fully be corrected) and just one E22 error (the most severe error which can be fully corrected) in all his hours of testing. The article dates back to 1990.
<edit> Armed with my vastly expanded knowledge of BLER, I'm now able to correct one of TerryO's misunderstandings:
Robert Harley got no significant uncorrectable errors in that article. So I'm not by any means alone. AccurateRip claims there have been 94million accurate rips of CDs using its database. Your misunderstanding here is not realising that even with significant BLER, the final data can be (and normally is) still perfect owing to the error correction.
<edit> Armed with my vastly expanded knowledge of BLER, I'm now able to correct one of TerryO's misunderstandings:
Robert Harley got no significant uncorrectable errors in that article. So I'm not by any means alone. AccurateRip claims there have been 94million accurate rips of CDs using its database. Your misunderstanding here is not realising that even with significant BLER, the final data can be (and normally is) still perfect owing to the error correction.
Last edited:
So if the data from (clean) CDs is coming off pretty much intact, I wonder why so many CD players sound harsh and edgy? In fact it's not just CDs - I was listening to an iPod playing system the other day at a show and that had the same brash midrange that irritated my ears. Strangely I did not hear a nasty sounding DVD that day - it was just CDs and iPods that sounded harsh. Actually there was an LP that sounded the same - harsh and irritating, on closer inspection they had put a digital processor in the signal chain. Odd though - as I've done the same and not noticed any harshness 😉
Of course cheap non-linear class B transistor amps with bags of feedback to multiply the harmonics probably did not help or may even have been wholly responsible! Maybe the people who sell this stuff think that's how it's supposed to sound..
Perhaps it's also more down to cheap DACs and or timing jitter.
When ripping my CDs I did notice some rip a lot faster then others so it obviously varies - just like LPs. In fact a couple of old CDs had great difficulty ripping at all, I think the silvering was deteriorating with age. I also wonder about the quality of the first analog filter caps that an oversampling DAC feeds into - I doubt they are silver mica, some may even be ceramic.
My own system that takes pre-ripped .wav files, declips and upsamples to 24/88.2 never sounds harsh, I'm still trying to work out the 'exact' ingredients that make my system sound so much better than the usual consumer hi-fi that ranges from ear hurting brash to dull plonking dullness that makes up the average hi-fi store offerings. I'm sure the all tube class A single ended amplification has a big effect.
I suspect the traditional CD harshness is a combination of things that all add up. Lampizator's page here (link) is an interesting read, especially his description of how upsampling changes the sound, and he listens via tubes too.
Of course cheap non-linear class B transistor amps with bags of feedback to multiply the harmonics probably did not help or may even have been wholly responsible! Maybe the people who sell this stuff think that's how it's supposed to sound..
Perhaps it's also more down to cheap DACs and or timing jitter.
When ripping my CDs I did notice some rip a lot faster then others so it obviously varies - just like LPs. In fact a couple of old CDs had great difficulty ripping at all, I think the silvering was deteriorating with age. I also wonder about the quality of the first analog filter caps that an oversampling DAC feeds into - I doubt they are silver mica, some may even be ceramic.
My own system that takes pre-ripped .wav files, declips and upsamples to 24/88.2 never sounds harsh, I'm still trying to work out the 'exact' ingredients that make my system sound so much better than the usual consumer hi-fi that ranges from ear hurting brash to dull plonking dullness that makes up the average hi-fi store offerings. I'm sure the all tube class A single ended amplification has a big effect.
I suspect the traditional CD harshness is a combination of things that all add up. Lampizator's page here (link) is an interesting read, especially his description of how upsampling changes the sound, and he listens via tubes too.
Two more "perhaps," which I think are more likely:
1. Perhaps the recordings are harsh and edgy? Commercial recordings by and large suck.
2. Perhaps your system balance is set up to compensate for the transient blunting, high second harmonic, and non-flat frequency response of an analog source, and when it's faced with a flat, clean source, it "colors" it in an unpleasant direction? I have sometimes wondered if that's why the Lampizator distortion boxes seem so popular...
1> One track that stands out is 'Money' by Pink Floyd on CD. This recording sounds pretty sweet in my system, the worst I ever heard it was at a Hi-Fi show with £38k of Chord dual monoblock amplification and about the same cost of tall modular french speakers (cannot recall the name). It was harsh, strident, devoid of bass and in fact took me a while to recognise it as the same track. But there it was, being torn apart and murdered by over £70k of 'hi-fi'.
2> Although I prefer good LPs for the special listen, most of the sound I play comes off a hard disk, through a declipper (which doesn't tend to touch much Floyd actually), then via optics into a Behringer Ultramatch to upsample to 24/88.2. From the DAC it then goes straight from the chip's output pin into my SE amp and into my Usher speakers. This flat, clean source does show EQ variations from track to track - some sound muddy, some nice and clear, but at no point do they ever sound harsh. In fact well recorded stuff like Moby, Melua, Shadows, Alisha's Attic, Floyd all sound exceptional - so I do not think it can be my system.
LPs through the same system (plus a tube phono pre!) sound equally sweet: but with more dynamics and a little more believable midrange.
I find brass instruments like trumpets or cymbals to sound quite harsh and sometimes even painful live and un-amplified. Not surprisingly when trying to produce a cymbal-like sound on an analogue synth the starting point tends to be a number of pulse wave oscillators set to unrelated fundamental frequencies or at a push a white noise generator could do. Harshness with these instruments is part and parcel of the real sound.
Thus I'd say (and in the process repeating myself) HD digital is more accurate than analogue but analogue sounds 'nicer'.
CD IMO is not particularly more truthful than vinyl, it just falls short in different, less nice ways. Mostly because its inability to produce 10k or so pulse waves, the harshness of the real instrument is there but lacking in 'body'. Analogue gets the body right but lacks the necessary hardness. We really need both and 24/96 would do the trick I think.
Thus I'd say (and in the process repeating myself) HD digital is more accurate than analogue but analogue sounds 'nicer'.
CD IMO is not particularly more truthful than vinyl, it just falls short in different, less nice ways. Mostly because its inability to produce 10k or so pulse waves, the harshness of the real instrument is there but lacking in 'body'. Analogue gets the body right but lacks the necessary hardness. We really need both and 24/96 would do the trick I think.
Poor design I'd say. I'd venture that most designers of CD players don't understand the finer details of grounding, decoupling and layout. If the designer has an analogue background he probably doesn't understand RF issues, and if he's a digital engineer he won't understand analog. CD players are an interesting mix of digital and analog circuits, presenting challenges from both sides.
I haven't found jitter to be a major source of harshness. Stray RF though most certainly is, particularly when it gets into opamps not designed for it.
Yes, I'd suggest that use of valves is most likely the reason your setup doesn't sound harsh.
Yeah, I've heard a LOT of expensive gear that made me wonder, "What were they THINKING?"
That track off the MFSL "gold" CD sounds pretty excellent on my speakers.
That is definitely NOT my experience. Reduce "Jitter"and the soundstage improves, you lose the hard edge of many recordings, and female voices typically sound sweeter.You can also then turn up the volume another notch or 2 without the sound being fatigueing. In fact, a well recorded CD can sound very analogue like. Actually, the big surprise is the improvement at the low end !
Good night Gents,
It's approaching midnight Downunder.
SandyK
- Status
- Not open for further replies.