-
People who believe that they can hear differences between 16-bit/24-bit or even 44.1kHz/96kHz+ sampling rates fall under these categories:
1. They're wrong and their ABX method is flawed, or they lied about doing ABX because doing a proper ABX is too much work (sorry!)
2. Their hardware/OS is interfering with the audio in a way that makes the output quality vary depending on source parameters (this could be caused by crappy output filters, crappy resamplers, unwanted signal processing somewhere in the chain, interferences caused by poor hardware implementation, etc)
3. In the case of 96kHz+ content, sometimes the ultrasonic information interferes with amplifiers/speakers causing additional issues which peoples' minds can intepret as "different" and "better"
4. "High resolution" content is often mastered differently, so you're hearing an improvement in the mastering process, not an improvement because of the high resolution itself
If you want to be scientific about it, you have to at least look at the output FFT of a test signal at all those test parameters to ensure that your hardware/software isn't messing it up. Check noise floor, check for harmonics, check for unwanted signals and possible intermodulation products, check square wave response to make sure nothing is oscillating (particularly relevant to DIY/modded hardware). The checks should be done at the final outputs (ie, speaker amp or headphones amp output). Do a white noise test and look at an averaged FFT to see if the linearity changes or if funky filters get switched in at different sampling rates, sometimes that's a thing..
In my experience, things that contribute to audio quality, sorted from highest contributing to lowest contibuting:
1. Speakers and room size/acoustics (together, this is by far the biggest factor)
2. Actual source quality (recording/mastering quality, format compression if any, bit-accurate output etc)
3. Basic issues (ground loops, bad power, interference between components - such as from switching PSUs etc)
4. Amplifier
5. DAC quality (power filtering, power regulation, output coupling, opamps/output stages, main clock oscillator phase noise, complexity of clock tree, dac chip jitter sensitivity, choice of dac chip)
6. Source path (ideally you want to avoid things like SPDIF that need clock recovery or cheap USB converters that carry really dirty power into the signal path and/or produce a lot of jitter because of a crappy clock... stick with I2S all the way if you can, since that's what the DAC chips will use). Some hardware generates I2S with microcontrollers or FPGAs, and is asynchronous to, and isolated from, the data input. Beauty.
1369. Your SD card brand (make sure to only use Sony's low-noise SD cards, or all this stuff will be in vain.
I'm a compulsive ABXer and anti-BS hardware dev person. I love proving myself/being proven wrong. I'm lucky enough to be able to hear to ~22kHz. :v I went to music school as kid. I have no friends because I science too much. If you can't trust me, then who can you trust...
-
Now I am use to the face most Americans go through life with suspension of disbelief hard wired in the on position. So it is quite normal to see snake oil, oxygen free copper, and global warming as the norm. All one need do is use the word SCIENCE! and all critical thinking gets turned off, brain in stand by mode....
PT Barnum laughs loud...
Remember if it is SONY it has to be PHONY. -
People who believe that they can hear differences between 16-bit/24-bit or even 44.1kHz/96kHz+ sampling rates fall under these categories:
1. They're wrong and their ABX method is flawed, or they lied about doing ABX because doing a proper ABX is too much work (sorry!)
2. Their hardware/OS is interfering with the audio in a way that makes the output quality vary depending on source parameters (this could be caused by crappy output filters, crappy resamplers, unwanted signal processing somewhere in the chain, interferences caused by poor hardware implementation, etc)
3. In the case of 96kHz+ content, sometimes the ultrasonic information interferes with amplifiers/speakers causing additional issues which peoples' minds can intepret as "different" and "better"
4. "High resolution" content is often mastered differently, so you're hearing an improvement in the mastering process, not an improvement because of the high resolution itself
If you want to be scientific about it, you have to at least look at the output FFT of a test signal at all those test parameters to ensure that your hardware/software isn't messing it up. Check noise floor, check for harmonics, check for unwanted signals and possible intermodulation products, check square wave response to make sure nothing is oscillating (particularly relevant to DIY/modded hardware). The checks should be done at the final outputs (ie, speaker amp or headphones amp output). Do a white noise test and look at an averaged FFT to see if the linearity changes or if funky filters get switched in at different sampling rates, sometimes that's a thing..
In my experience, things that contribute to audio quality, sorted from highest contributing to lowest contibuting:
1. Speakers and room size/acoustics (together, this is by far the biggest factor)
2. Actual source quality (recording/mastering quality, format compression if any, bit-accurate output etc)
3. Basic issues (ground loops, bad power, interference between components - such as from switching PSUs etc)
4. Amplifier
5. DAC quality (power filtering, power regulation, output coupling, opamps/output stages, main clock oscillator phase noise, complexity of clock tree, dac chip jitter sensitivity, choice of dac chip)
6. Source path (ideally you want to avoid things like SPDIF that need clock recovery or cheap USB converters that carry really dirty power into the signal path and/or produce a lot of jitter because of a crappy clock... stick with I2S all the way if you can, since that's what the DAC chips will use). Some hardware generates I2S with microcontrollers or FPGAs, and is asynchronous to, and isolated from, the data input. Beauty.
1369. Your SD card brand (make sure to only use Sony's low-noise SD cards, or all this stuff will be in vain.
I'm a compulsive ABXer and anti-BS hardware dev person. I love proving myself/being proven wrong. I'm lucky enough to be able to hear to ~22kHz. :v I went to music school as kid. I have no friends because I science too much. If you can't trust me, then who can you trust...
I think i do hear the difference between 24 and 16bits, but i don't hear any difference between 41khz and 48khz or 96khz. But for my test i use crappy recording, crappy integrated Realtek ALC1200 and crappy speakers/headphones. But in my crappy equipment 24 bits work. -
Sony is stupid. Why? Because if you look at the costs of the audiophile products sold to these idiots, like $600 for a 1 meter power cable -
http://www.lessloss.com/dfpc-series-p-213.html
or $1,600 for a 6 ft speaker cable with mono crystal pure copper strands and vibration absorbing nanoparticles (yes they really say that) -
http://www.lessloss.com/anchorwave-interconnects-and-speaker-cables-p-205.html
Sony is stupid because they could charge a whole lot more for the card than they are.
-
Sony is stupid. Why? Because if you look at the costs of the audiophile products sold to these idiots, like $600 for a 1 meter power cable -
http://www.lessloss.com/dfpc-series-p-213.html
or $1,600 for a 6 ft speaker cable with mono crystal pure copper strands and vibration absorbing nanoparticles (yes they really say that) -
http://www.lessloss.com/anchorwave-interconnects-and-speaker-cables-p-205.html
Sony is stupid because they could charge a whole lot more for the card than they are.
I haven't seen any evidence people actually buy that crap.
However;
People are stupid, someone once said there are two things in the Universe that have no limit;
Love
and
Stupidity. -
It is, although it's worth noting that the SPI bus itself has to be carefully managed because otherwise you are going to start seeing high frequency switching noise. Maybe it has better termination than the average card, which would certainly reduce emissions through the contacts.
While I agree with most of the problems you described, I have to disagree with the memory card SPI related noise. Changing only the memory card in a system will not have any difference in a properly designed system.
Even if you disagree that 24 bit playback is worth while, surely you must accept that if the goal is to implement it then very small amounts of electrical noise will quickly raise the system noise floor far above where the digital one is.
The frequency the SPI read bursts happen is below the loop frequency of a power supply. If the power supply rails are properly routed on a board, the burst will be inaduable. Also, the only way to decrease the noise on the Mhz++ level, which they claim they did is to decrease the loading of the digital lines. It is possible that they did this, but most of the loading will be due to the PCB traces anyway. You can argue, that if you decrease the rise time of the signals (not changing the baudrate), that will generate less high frequency noise, but you dont change the microcontroller's rise time, so the total effect is questionable to say the least. Shielding the internals should be by the host, not the card, and the host should have proper bypassing of the power supplies.
It is a snake oil device, even if the basis they used (if there is any) for the changes are scientific facts. It is like putting spikes below your DAC.
Spikes are good for mechanical dampening.
Mechanical forces affect electronics, reference voltage for example (I can give you a linear technology appnote to back this).
But no, putting spikes below a DAC will not have an audible effect.
-
4. "High resolution" content is often mastered differently, so you're hearing an improvement in the mastering process, not an improvement because of the high resolution itself
No kidding.
You remember those SACD disks - 24bit, 192kHz or whatever they were...? They were made in layers - a normal CD at the bottom then the "SACD" layer underneath that.
Well, about ten years ago I dropped one of my favorite CDs and scratched it enough to make it skip.
(Remember when they told use we could drill holes in them? Was that SONY as well?)
So... I went out and bought another copy. But the new copy was crap. It had been "remastered" or something and was all horribly distorted (record companies take real pride in their work, they often used to record CDs onto 48kHz DAT tapes then send them off the duplicators to make a new batch of CDs)
So what could I do? I checked on eBay and saw an "MFSL" version of the CD. MFSL are people who take old records and remaster then into 24 bit recordings.
http://www.mofi.com/
They pride themselves on the quality of their work so I took the plunge and bought the CD. Hopefully I'd get a decent version of the record.
When it arrived it was absolutely awful. It was low pass filtered - no treble left at all.
I assume that SACD players have a button on them to switch between "16-bit CD" and "24-bit SACD" and SACDs are being deliberately mastered so that audiophools canpress "24 bit" and marvel at all the extra treble. (ie. the treble they've deliberately removed on the 16-bit version).
I actually made a comparison of the SACD sound vs. normal CD. There's a GIF below of a waveform with a hissy-whooshy sound in it. Compare the two. It's not even close.
What did I do in the end? I downloaded a FLAC file off the Pirate Bay of course. I own two CDs and a SACD of the music so I figure they can't complain.
(And it's crap like this that made me the pirate I am today...they bring it on themselves)In my experience, things that contribute to audio quality, sorted from highest contributing to lowest contibuting:
1. Speakers and room size/acoustics (together, this is by far the biggest factor)
Yep. The single biggest thing you can do to improve your HiFi is to move the speakers around, preferably away from walls (and especially corners). And it's free to do (unless it annoys the wife...which it usually does)
The second thing? Spend more on your speakers than the amplifier. It goes against everything that they say in magazines but it really works. I've seen people buy expensive amplifiers and attach total cornflake packets to them.
Think though - electronics is mostly a done deal. An EEVBLOG type can build you a damn good amp for $150. It might not look pretty but it'll sound as good as almost anything in the HiFi store. It's just a decent transformer plus a few MOSFETs and other stuff.
Speakers? They're physical objects. They depend a lot on quality materials and good construction. They cost a lot more than $150 to build them properly.
-
Remember as well, from JLH IIRC, that a $150 speaker might only have has $15 spent on the drivers inside. An amplifier typically has a plateau above which the improvements are logarithmic. You get cheap, then double the price and get adequate, double again to good and then every doubling above that gives a very small improvement if at all.
-
3. In the case of 96kHz+ content, sometimes the ultrasonic information interferes with amplifiers/speakers causing additional issues which peoples' minds can intepret as "different" and "better"
HD audio is low pass filtered to remove content above the audible range anyway. The reason for going to 96KHz is to reduce aliasing of high frequency signals, not to allow higher frequencies.Quote4. "High resolution" content is often mastered differently, so you're hearing an improvement in the mastering process, not an improvement because of the high resolution itself
Indeed, but that alone is reason enough to welcome HD audio. One of the reasons why SACD and DVD audio are popular is that they require the mastering process to meet certain requirements set out by Dolby, which includes things like the minimum average dynamic range that limits compression. Same reason people buy old second hand CDs from the 80s, same reason they buy vinyl in fact. Old CDs were properly mastered, vinyl simply can't support the amount of compression that modern CDs can due to limitations of the format.
Hmm I think it's more correct to say that HD audio is lowpass filtered SOMETIMES. A lot of the content I've looked at extended to ~50kHz, some had 24kHz cutoff but not many. SACD will have your characteristic ultrasonic peak as well, which is tamed, but never completely filtered. Aliasing isn't an issue as long as you stay within nyquist. If you play vinyl, then again it will depend on the source. Some 30ips R2R tape machines can do 40kHz+, which is what the masters are recorded on, and what the vinyl is pressed from. Some high quality MC cartridges can do 10Hz - 60kHz within -3dB. Phono stages are supposed to have ultrasonic filters, but they're all different and wildly inconsistent. I don't actually listen to vinyl, but I've developed custom phono stages for clients.
I completely agree that older music was mastered better. Studio time was very expensive, and great attention was paid to preserve signal integrity in the analog world - simply because a lot of things could ruin it. These days, it's all digital filters.. but.. how good are they? Who characterizes them? How do they play together? How many people have truly flat monitors? How many people know what "flat" sound should be like? How many people strive for realism and a sense of space, instead of "sound that pops" lol? It took digital reverbs forever to catch up to a real life plate reverb, just due to computational complexity involved for realtime operation. :/
Good primers to understand digital/analog with respect to audio:
http://xiph.org/video/vid1.shtml
http://xiph.org/video/vid2.shtml -
Speakers? They're physical objects. They depend a lot on quality materials and good construction. They cost a lot more than $150 to build them properly.
Definitely so. There is an easy test you can do for speakers which works quite well - knock on them. If they sound hollow or resonate, they won't sound good.
Lesson of the day: SPEAKERS ARE WATERMELONS!
Hahah, it's funny but true. Speaker building is an engineering art, much like building a musical instrument. Just look at the engineering that goes into some modern high-end brands like Magico.
I'm a big fan of room correction as well, there are some great devices out there like the Accuphase DG-58 which makes it really painless but far from cheap, but well.. there are many other nice (and free) software solutions as well, which even includes phase correction... -
I think i do hear the difference between 24 and 16bits, but i don't hear any difference between 41khz and 48khz or 96khz. But for my test i use crappy recording, crappy integrated Realtek ALC1200 and crappy speakers/headphones. But in my crappy equipment 24 bits work.
Generally speaking, you need a really high quality setup to hear fine differences in sound, as well as training to do so. In your case, it sounds like it's more of a problem specific to that ALC1200 chipset which the 24-bit source overcomes. I use an ALC898 on a Rampage IV motherboard as a "baseline" reference actually, because sometimes I want to remind myself of what a "normal" dac/preamp sounds like, so that I don't let placebo take over when it comes to making design improvements. A good goal for any person, even if they're not an "audiophile", is realism. Wouldn't it be awesome if sounds from your headphones were real enough to trigger a subconscious startling reaction? Imagine if you couldn't distinguish between real sounds and what you hear on headphones. Imagine playing a horror game like that on the Rift. Well, we're nowhere near that point, sadly. However, I think that's a great goal to aspire to in the audio world...
-
Sony didn't invent the Walkman.
YES, they did.
Here's a catalog from Sony, year 1971: http://wegavision.pytalhost.com/Sony/1971/011.jpg
Look at the portable cassette player, TC-40. That's basically a Walkman. Sony just didn't have the idea to actually call it a Walkman before 1979 when they decided to market it as a portable listening device for music.QuoteIt was a German who invented it.
No, he didn't. If he had, he would have easily won the case in court which he didn't.QuoteSony's first Walkman prototype was nearly a one to one copy of the german design.
No, it wasn't. The TPS-L2 was basically one of their previous portable cassette players/dictation devices with the recording functionality removed. This is why the TPS-L2 still has a microphone built in which they marketed as a "attention" feature which would mute down the music and allow you to listen to your environment when your attention is needed.QuoteSony lost the lawsuit after nearly 30 years.
No, they didn't. They settled the case outside of court because they were tired fighting with that guy as he wasn't giving up. And they could only do it after Akio Morita passed away who had the original idea to market Sony's portable cassette devices as portable audio listen devices, later called Walkman and would have NEVER agreed to give in. He invented the Walkman the way people later used it.
Endlessly repeating that myth that Pawel invented the Walkman doesn't make it any more true. Sony had portable cassette players already in 1970/71 so anything that Pawel had was already preceeded by prior art.
Adrian -
I think i do hear the difference between 24 and 16bits, but i don't hear any difference between 41khz and 48khz or 96khz.
You might want to watch this and understand why it is mathematically not possible to hear a difference between 16 and 24 bits sampling:
http://xiph.org/video/vid2.shtml
The only thing that is actually audible is a sampling frequency which is too low and therefore the low-pass filter at Nyquist frequency / 2 has to be very sharp which is generally hard to implement which is why your resulting audio signal will contain aliasing noise.
Adrian -
The "rounding" you are thinking of is quantization, but nothing in the design of the memory card has anything to do with quantization. Selling this memory card for superior sound quality is like selling sheet music printed on canvas vs. regular paper, for superior playback by a violinist reading music. It's actually much worse than that, because a violinist could miss a note - but a digital music player isn't missing 1's and 0's from a non-audiophile card.
Question: do you think 24 bit recording is pointless? Have you ever compared it directly with a 16 bit recording in a double blind test? Having said that, such a test is of somewhat limited use because 16 bit mandates different mastering techniques and parameters, but even so...
No, I think 24 bit recording is very valuable. The range of music is around 80db (max - most of it is way, way less), but recording in 24 bit allows the audio engineer to be a lot less careful about setting their levels during the recording process. It also gives them more headroom during the mastering/adjustment process.
For delivery to users, 24 bit recording is completely pointless. The only difference is a lower noise floor, but considering 16-bit can already hit the threshold of pain when the noise floor is adjusted to anechoic chamber levels, lowering the noise floor (or increasing the top end) has no effect.
Yes, I've done plenty of double blind tests - there are threads over on Hydrogen Audio and Head-Fi where dozens of people have done DBT's and not a single person has ever been able to tell the difference with all other things being equal - no human can. Any perceived differences come in the mastering process (which can make a HUGE difference). -
People are stupid, someone once said there are two things in the Universe that have no limit;
Einstein said: "Only two things are infinite: the universe and human stupidity. But I am not sure about the universe."
Love
and
Stupidity. -
It's a shame he only made two videos. Hopefully he will make more someday.
You might want to watch this and understand why it is mathematically not possible to hear a difference between 16 and 24 bits sampling:
http://xiph.org/video/vid2.shtml
Interesting video link. Thanks for that. -
Hey Dave, that SD card is just the tip of the iceberg. Try this: http://www.audioquest.com/ethernet/diamond
Ethernet cable; 5500$ for 8 meters.
Want a power cable for 7000$? This is for you. http://www.audioquest.com/power-cables/wel-signature
That network cable that you're refering?
Here is some more audiophoolery that grind my gears:
http://www.geek.com/chips/this-ethernet-cable-costs-10000-1615326/
They claim: "You don’t want some crappy unbiased cable, because it’ll “slow down parts of the signal differently” and that’s “a big problem for very time-sensitive multi-octave audio.”
THIS IS A NETWORK CABLE!!! Can someone try to explain me how the hell audio quality, that comes from a digital stream, carried on UDP packets, and probably encoded by a lossy MP3 algorithm, will affected by the cable???
Let me guess: definitely not the lossy MP3 algorithm, or UDP protocol... Oh I know! It must have unicorn fart powder!!
-
This SD card is gold and white!
End of story. -
No, I think 24 bit recording is very valuable. The range of music is around 80db (max - most of it is way, way less), but recording in 24 bit allows the audio engineer to be a lot less careful about setting their levels during the recording process. It also gives them more headroom during the mastering/adjustment process.
Recording at 96k samples/second is also valuable, because it pushes the anti-alias filters so far from the audio band that many generations of manipulation still leave the top edge of the audio band clean. Its totally pointless for distribution to an end user, though. People seem to struggle with the the difference between working data and final results. -
I just gotta say, that of all your videos
BruceDave, this one had me laughing the most!
-
There is something similar going on in the classical music world, where famous performers and music critics are convinced that old famous string instruments sound better than new ones. (A Guarneri or Stradivari violin sells for millions of dollars at auctions).
Partially true. In the case of piano, the iron frame changes the sound. The tunning was also different.
This was thoroughly debunked in double blind experiments, but the myth lives on.
-
Perhaps, but the piano is classed as a keyboard instrument. You're muddying the waters, pitch tuning doesn't influence the sound quality of the instrument.
-
Strings are producing sound. Not the keyboard
-
I'm not sure what you're trying to prove. The sound quality of a violin is in the workmanship of the sound box construction, not the strings.
There is no controversy about pianos, a modern Steinway is as good as an old one. -
I don't want to prove anything. Piano is a very specific instrument where strings, as well as hammers and soundboard play their own role. Some old compositions may sound better on old pianos than new ones. But at the end, everything is subjective.