Why do records sound so much better than digital?

[levimax]

So is it a "flat transfer" of the original master tape what we want to hear or do we want to hear what the mastering engineer did to the original master tape back in the day? While still hit or miss if you get an original pressings of an LP's signed in the dead wax by the original mastering engineer at least you know he was working with original master tapes and made adjustments he thought sounded good. While not necessarily the best sounding versions at least the signed original pressings are what people heard back in the day which I find interesting and one reason I enjoy collecting and listening to LP's.

 

[rdenney]

I think they were, at least that was my impression doing needledrops for "Music from the Hearts of Space". I mentioned Denon's very early [1970s] digital recordings. My sonic impression of early, subpar digital was that sounds were too soft, there was an issue with resolution. I could compare copies rendered in three different formats---Tascam 32 reel to reel at 71/2 IPS, high quality cassette and the Sony 501. The 14-bit digital transfers had no additional noise but could seem to blur the sound more than 16 bits. Overall best of the three [sonically] were the cassettes, using metal formulation and no Dolby. Pretty much matched the noise floor of the best LPs, like French Harmonia Mundi. The reel-to-reel machine was both noisy and more distorted. The digital had the least noise but seemed to suck the life out of the sound. As I was still reading TAS that could have been all in my mind. But the Technics DA-10 that followed sounded a lot more like a bog-standard CD player. I was also hearing distortions with CD players back then that I might not hear now.

And yes, I am making your point, though my other point would be that a lot of audio myths started a long time ago, when there were substantial differences in the performance of ADCs, DACs and suchlike. 30 years is aeons for the development of digital anything, and modern digital is functionally transparent. But back in 1992? Not so much.

It was around 1992 when I was making live recordings using HiFi VHS. I did play in a concert in that time range that was recorded using a Sony digital interface to some kind of tape machine--I was busy with musician duties and didn't take note of the details. So, yes, that was about when the first D of DDD appeared, I suppose, in the hands of mortals. Don't know what the bit depth was, but the recording was excellent. I probably started recording digitally maybe five years later.

So, I got curious. I had a Soundblaster 16 for an ISA-bus PC clone back in the day. Apparently, those came out in 1992, and were the first to record line-level inputs in "CD-quality" (16-bit 44.1 KHz sampling to PCM). I read an old test that measured them at -70 dB THD (0.03%) and about that for S/N. So, the first popular consumer 16-bit sound-card interface still had better performance than LPs. So, 30 years--just barely. Five years later 24-bit sampling was common. My Yamaha pro digital parametric equalizer from around 2000 is still excellent and used 20-bit 48KHz internal encoding and decoding.

(It's actually kinda hard to find out how that old hardware worked for sound recording--the keepers of that information are into gaming and thus uninterested in the gozintas versus the gozouttas.)

Rick "built a 25-foot RCA audio cable to feed from the stereo to the computer in the middle 90's, when playing with MIDI keyboards and so on" Denney

 

[Audiofire]

So, I got curious. I had a Soundblaster 16 for an ISA-bus PC clone back in the day. Apparently, those came out in 1992, and were the first to record line-level inputs in "CD-quality" (16-bit 44.1 KHz sampling to PCM). I read an old test that measured them at -70 dB THD (0.03%) and about that for S/N. So, the first popular consumer 16-bit sound-card interface still had better performance than LPs. So, 30 years--just barely. Five years later 24-bit sampling was common. My Yamaha pro digital parametric equalizer from around 2000 is still excellent and used 20-bit 48KHz internal encoding and decoding.

I did measure my Asus Z97-Pro Gamer motherboard, but there was some difference in level that didn't quite work for spectrograms. If someone knows how to match the level for looping an audio file, I would like to know.

I measured again now that you mentioned it and there was about a half percent distortion in REW (if I set it up right). But the background noise is audible on high sampling rates, the noise from mouse movement is also audible in headphones.

 

[rdenney]

If you can hear the noise from mouse movement, you are probably mixing in the computer's microphone, which would definitely add a loud noise floor.

REW will let you adjust levels if you follow its procedure. I use an external sound interface that has analog level controls for using REW (a Presonus Studio 24c), so I'm not sure how to do that in the OS. Maybe someone with more skills using REW can help.

Rick "who measured the Presonus at -100 dB noise floor in loopback mode" Denney

 

[Bleib]

Hearing the mouse movement can sometimes be fixed by changing the mouse to a USB2 port instead of 3.x

 

[Audiofire]

Hmm... Can't hear mouse movement or background noise on the motherboard now. I want to sell it, so made some more tests. Could have something to do with that I was testing an old preamp and might have plugged the headphones into that one.

 

[CherylJosie]

Examples?

I could once hear the difference between live recordings on open-reel tape and live recordings on VHS HiFi. But I have never been able to hear the difference between ADCs. I have a Creative external sound card of some sort—I don’t recall the model. It tested about as poorly as any of them do. And I currently use a highly rated Benchmark ADC. Absent clipping one of the preamps or screwing up the gain structure, I cannot hear any difference between them when used for needledrops. Played through speakers or headphones, they sound just like what’s coming out of the cartridge.

I’m thinking that an ADC with 80 dB of S/N and dynamic range over the audible spectrum is going to be indistinguishable from one that provides 110 dB, when recording media that only provides 60 or 65 dB dripping wet, not including the -40 dB vinyl roar. Yes, it can be done poorly with an inappropriate reconstruction filter or a poorly set up gain structure. A needledrop is no easier than making a tape of an LP back in the day.

I doubt that any available sound card is so poor that it would be noticeable unless it added hum or was overdriven.

Now, the software in the computer may be performing all sorts of processing that will have audible effects, like overly aggressive click removal or whatever. I use Vinyl Studio, and some of its filters will leave audible artifacts when listening to the tail of a fade at extreme amplification. But one can leave those filters turned off.

But people rarely conduct the test for themselves, and just perpetuate their own biased perceptions.

Rick “uses the Benchmark because of its detailed gain control” Denney

Back in the early days the 1 bit delta-sigma converter was all the rage, until somebody developed a theoretical explanation for why such converters always saturate and introduce distortion when dithered.:

http://www.essentialaudio.com/1-Bit-Is-Bad.pdf

Archiving and mixing in SACD format turns out to be suboptimal (and mathematically challenging too where the mixing/processing part is concerned)

Then when people started designing multi-bit delta-sigma converters they discovered that the multi bit nonlinear feedback path around the sampler exhibited unpredictable chaotic behavior where sampling a sine wave could result in spurious tones being generated that shift around in frequency and amplitude depending on the schedule offset of the input signal. It’s not harmonic distortion but rather chaotic behavior.

Depending on how many bits and what the order of the nonlinear feedback filter is, these suboptimal behaviors can be fine-tuned but never completely eliminated.

Converter design has come a long way since the audio CD format was first released. It’s no wonder that some people didn’t like early renderings. They were probably detecting audible artifacts, both in the digitization process and in the analog playback process.

 

[Audiofire]

Then when people started designing multi-bit delta-sigma converters they discovered that the multi bit nonlinear feedback path around the sampler exhibited unpredictable chaotic behavior where sampling a sine wave could result in spurious tones being generated that shift around in frequency and amplitude depending on the schedule offset of the input signal. It’s not harmonic distortion but rather chaotic behavior.

Depending on how many bits and what the order of the nonlinear feedback filter is, these suboptimal behaviors can be fine-tuned but never completely eliminated.

Converter design has come a long way since the audio CD format was first released. It’s no wonder that some people didn’t like early renderings. They were probably detecting audible artifacts, both in the digitization process and in the analog playback process.

I'm not sure this makes any difference with the good multibit delta-sigma converters like RME ADI-2, someone else might know more about it.

 

[dlaloum]

Back in the early days the 1 bit delta-sigma converter was all the rage, until somebody developed a theoretical explanation for why such converters always saturate and introduce distortion when dithered.:

http://www.essentialaudio.com/1-Bit-Is-Bad.pdf

Archiving and mixing in SACD format turns out to be suboptimal (and mathematically challenging too where the mixing/processing part is concerned)

Then when people started designing multi-bit delta-sigma converters they discovered that the multi bit nonlinear feedback path around the sampler exhibited unpredictable chaotic behavior where sampling a sine wave could result in spurious tones being generated that shift around in frequency and amplitude depending on the schedule offset of the input signal. It’s not harmonic distortion but rather chaotic behavior.

Depending on how many bits and what the order of the nonlinear feedback filter is, these suboptimal behaviors can be fine-tuned but never completely eliminated.

Converter design has come a long way since the audio CD format was first released. It’s no wonder that some people didn’t like early renderings. They were probably detecting audible artifacts, both in the digitization process and in the analog playback process.

And yet, the best 14 bit players, were very good indeed...

 

[fpitas]

Purely anecdotal. But a friend who gets a lot of vinyl carefully ripped several LPs to lossless digital. Then we spent a lot of time A/Bing, without hearing much difference.

 

[stalepie2]

Maybe it's unnatural to hear the exact same sounds again and again ("perfect sound forever") and analog forms of music don't sound exactly the same, even if the record hasn't degraded much, maybe the tone arm drags a little too much one time and not the other? People seem interested in having lots of headphones in order to hear the same digital recordings a little differently, so that they stay fresh. So you could have digital audio that degrades over time, artificially, using Digital Rights Management (DRM), and you'd buy or stream an audio file and the more it's played the more noise accumulates. As an effect that's added in, like the Vinyl Sound Processor on the Sony A55 Walkman. But it could get worse over time with no way to fix it. Then the music would be more precious and people would care more about owning it and taking good care of it. Maybe there could be artificial ways of pretending to clean the file too to get rid of some of the dust. People enjoy games like Nintendogs and Tamagotchis and The Sims and stuff like that, where they have to take care of virtual creatures. NFTs could do this. The more they are viewed, it's like light exposure from the sun on oil paintings, and needs to be minimized. The more they're shared online, it could degrade, similar to how you're not supposed to take flash photography at a museum.

 

[Newman]

^ ^ ^

 

[antcollinet]

Maybe it's unnatural to hear the exact same sounds again and again ("perfect sound forever") and analog forms of music don't sound exactly the same, even if the record hasn't degraded much, maybe the tone arm drags a little too much one time and not the other? People seem interested in having lots of headphones in order to hear the same digital recordings a little differently, so that they stay fresh. So you could have digital audio that degrades over time, artificially, using Digital Rights Management (DRM), and you'd buy or stream an audio file and the more it's played the more noise accumulates. As an effect that's added in, like the Vinyl Sound Processor on the Sony A55 Walkman. But it could get worse over time with no way to fix it. Then the music would be more precious and people would care more about owning it and taking good care of it. Maybe there could be artificial ways of pretending to clean the file too to get rid of some of the dust. People enjoy games like Nintendogs and Tamagotchis and The Sims and stuff like that, where they have to take care of virtual creatures. NFTs could do this. The more they are viewed, it's like light exposure from the sun on oil paintings, and needs to be minimized. The more they're shared online, it could degrade, similar to how you're not supposed to take flash photography at a museum.

Utter nonsense.

Sorry.

 

[Audiofire]

Everything about vinyl records is more cumbersome and worse sound quality, so the level of deliberation required is much higher and that engages production engineers as well as end users.

 

[stalepie2]

Utter nonsense.

Sorry.

why?

 

[Newman]

Mate, I assumed you were pulling the collective leg.

 

[Peluvius]

The only reason I still enjoy a few select Vinyl albums is because the sound reminds me of the time in my life when I first discovered them (and other relevant associations). The colouration is familiar to me. As a broad generalisation Vinyl sounds "different" because it introduces additional sounds that were not present in the original recording. If you believe this is superior then that is your business however please don't suggest that Vinyl will more accurately reproduce the recording.

 

[Audiofire]

The only reason I still enjoy a few select Vinyl albums is because the sound reminds me of the time in my life when I first discovered them (and other relevant associations). The colouration is familiar to me. As a broad generalisation Vinyl sounds "different" because it introduces additional sounds that were not present in the original recording. If you believe this is superior then that is your business however please don't suggest that Vinyl will more accurately reproduce the recording.

Vinyl can sound more like the recording when a more natural dynamic range is preserved.

 

[Audiofire]

why?

Since vinyl is very limited for reproducing natural music, digital sounds more natural with post-production taken into consideration. The degradation in vinyl you talked about is precisely because it sounds even less natural with degradation. People might use different headphones in order to preserve the digital reproduction, not to degrade it. EQ would have a similar effect.

A DRM file can be recorded one time and saved as a digital file. The difference in quality won't be great or noticeable if similar quality is used for saving the digital file.

 

[krabapple]

Genuine question, how do we know that those audiophile remasters are using the original tapes? Is there a way to verify it, or we've got to take the label's word for it?

IF the remaster is on a high-rez format, a sharp cutoff at 22 kHz in a spectral analysis would be an indicator.

Otherwise, yeah, you just have to take their word for it.