How about that sound

[Theo Verelst]

I recall clearly that some people in the time of the first CDs (I was teenager) were comvinced that one way or another: digital was equivalent with perfection!

 

Now, while digital recording might in some sense be able to get close to that ideal, digital processing, and digital to analog conversion usually are quite far from perfect.

 

Most modern plugins to me don't sound like a streamlined yet with good powe control, but more like these kinds of contraptions:

 

[video:youtube]Mm-cIUEzKr4

 

I've worked on making digital processing that can sound "un-digital" for lack of a better word, an preparation of digital signals to come out of normal DACs more properly prepared, which honestly is a lonely path.

 

Theo V

[Notes_Norton]

I recall clearly that some people in the time of the first CDs (I was teenager) were comvinced that one way or another: digital was equivalent with perfection!<...snip...>

 

Theo V

 

That was a lie perpetrated by the recording industry. The resolution was way too low.

 

On the 25th anniversary of the CD, I read an article in the trade magazines by one of the inventors of the CD. He admitted that there were severe quantization errors in the format which reproduced harmonic frequencies that were manufactured by the AD and DA conversion process. CDs traded the distortion caused by analog records, mostly noise, for a different kind of distortion - a change in the tone of the instrument or voice by adding harmonic frequencies.

 

They sold the CD to the record companies by telling them it was much cheaper for them to manufacture, therefore increasing the corporate profits.

 

The recording industry, knowing that they wanted the profits but couldn't admit why they were pushing CDs, decided to coin the phrase "CD Quality".

 

Now I'm a firm believer that tone doesn't matter nearly as much as the musical content. Since my youth people bought 45RPM records, 8 track tapes, cassettes, and mp3s sometimes listening through cheap ear-buds or tinny phone speakers. And what about compressed audio streaming to a cell phone?

 

Case in point: The higher resolution SACD which sounded much better to my ears never made it The public didn't care.

 

Understanding all that, the distortion created by both old and new recording processes doesn't worry me too much. I know that my listening tastes are not like the tastes of the general public. I listen with trained musician ears.

 

I find I like the tone of a well recorded vinyl record on a nice system better than the tone of a well recorded CD, but I don't like the surface noise of the vinyl record. It's a trade off. I listen to CDs more often because they are more convenient.

 

I own a lot of LPs and a lot of CDs, and for my car, I have over 10,000 songs ripped from my own collection in 128 mp3 format.

 

There is no perfect recording method. Everything introduces some kind of distortion. As my dad used to say "pick your poison".

 

I just enjoy the music I like and the format depends on what I'm listening to and where I'm doing the listening.

 

Insights and incites by Notes

 

 

 

[Anderton]

I recall clearly that some people in the time of the first CDs (I was teenager) were convinced that one way or another: digital was equivalent with perfection!<...snip...>

 

Theo V

 

That was a lie perpetrated by the recording industry. The resolution was way too low.

 

Now hold on there a second...are you telling me that marketing departments don't tell the truth? I bet you don't believe in the Easter bunny, either.

 

One of the biggest problems with early CDs wasn't the CDs, it was the converters in the players. A lot of consumer CD players used 12-bit converters.

 

The main problem with digital is that distortion increases as the volume decreases, which goes against how we normally experience sound in the real world. If I have the math right, which certainly isn't guaranteed, with a 16-bit format you're listening with 12-bit resolution at -24 dB. Of course we're mostly recording at 24 bits in our home studios, but we still have the delivery medium issue.

 

I agree that SACD indeed sounded better, and I'm not a wine-tasting, self-proclaimed "golden ears" guy when it comes to music. Whether it was because of the one-bit encoding technology or the ability to avoid brickwall output filters is a subject of debate. Also remember that a lot of SACD releases were redone to be optimized for the medium.

 

So...yes, all forms of audio reproduction have their issues. It is what it is, and yes, ultimately the musical performance is what matters anyway

[Mike Rivers]

I recall clearly that some people in the time of the first CDs (I was teenager) were comvinced that one way or another: digital was equivalent with perfection!<...snip...>

 

Theo V

 

That was a lie perpetrated by the recording industry. The resolution was way too low.

 

On the 25th anniversary of the CD, I read an article in the trade magazines by one of the inventors of the CD. He admitted that there were severe quantization errors in the format which reproduced harmonic frequencies that were manufactured by the AD and DA conversion process.

 

It's true that the original CDs weren't quite "perfect," but it wasn't a matter of the resolution being too low. Indeed there was distortion, but it was due to the converters of the period. Non-linearity of the conversion between analog to digital and back again was one problem, noise was another, with a 16-bit converter having 12 or so bits representing the best accuracy they could, with the remaining bits being noise. Further, the "brick wall" anti-alias filter on the A-to-D input and reconstruction filter on the D-to-A output had massive phase shift, causing smearing and harmonics getting out of sync with their fundamentals.

 

If you were to record audio at 44.1 kHz, 16-bit resolution using today's hardware you or the quoted inventor of the CD would be happy with the sound.

 

[Anderton]

Non-linearity of the conversion between analog to digital and back again was one problem, noise was another, with a 16-bit converter having 12 or so bits representing the best accuracy they could, with the remaining bits being noise.

 

To be clear...when I mentioned using 12-bit converters being a problem, I didn't mean you could get only 12 useable bits from a 16-bit converter, I meant 12-bit converters. So they only had, what, about 10 real bits?

 

[Mike Rivers]

To be clear...when I mentioned using 12-bit converters being a problem, I didn't mean you could get only 12 useable bits from a 16-bit converter, I meant 12-bit converters. So they only had, what, about 10 real bits?

 

As much as I wanted to start playing with recording audio on a computer, I just couldn't bring my self to getting an 8-bit Soundblaster card, but I sprung for a second hand Turtle Beach Maui card that had 16-bit converterss, and that was really hot stuff for the day.

 

Although I never measured it, I suspect that 12-bit converters had fewer marketing bits than the 16-bit converters of the day. And 16-bit was always the CD standard. 10 bits gets you about the same dynamic range as analog tape, but each had its own set of non-linearity above the noise floor. The spectral content of the distortion (that is, what's added to the signal besides random noise) is different and there are good reasons why, as long as we have to live with distortion, analog distortion sounds more pleasant to our ears than digital distortion, even though the digital distortion may be at a lower amplitude than analog. Today's converters, however, are another story. Now they're selling us 32-bit converters. Marketing bit-count has always been that it's not how many bits before you lose accuracy, it's how many bits the word length that comes out is.

 

What was the name of that 8-bit sampler keyboard that became quite popular because it was the first sampler that working musicians could afford?

 

[Notes_Norton]

I remember the Turtle Beach cards, they were awesome for their day.

 

But if one of the inventors on the CD team publishes that the bit rate was too slow and caused severe quantization errors, I figure he knows a lot more about the subject than I do.

 

And yes, DA converters have gotten better so CDs don't sound as tinny as they used to.

 

But I repeat, if the general public really cared about audio quality, the SACD wouldn't have died wasted death.

 

It seems the public likes video resolution more than audio judging from the screens I see in the stores.

 

Insights and incites by Notes

[Anderton]

What was the name of that 8-bit sampler keyboard that became quite popular because it was the first sampler that working musicians could afford?

 

You're probably thinking of E-Mu's Emulator II. I did the manual for it and amazingly, still have it but haven't booted it up in a while. They actually got 12-bit operation through companding.

 

[Theo Verelst]

The problem is that there's a mathematical "reconstruction theorem" that states you need a near infinite resampling/oversampling filter in order to get close to retrieving the samples that you've made as in the original analogue sound quality.

 

This receives little attention, but it is next to impossible to make an existing Digital to Analog Converter sound neutral such as to arrive at high fidelity sound where dimensions and sound placement accuracy are really good.

 

So it's a technical or scientific problem to deal with all existing audio DACs in such a fashion that they sound comparable to a good tape. You can just ignore this problem, and say the reconstruction (theoretical EE terminology) bug is actually snug in in the rug of the New Perfection, but it is an error to think your DAW can playback tracks with HiFi perfection, and just about every digital playback of anything proves this.

 

Then there is the digital *processing*. That's worse in the sense that no regular audio engineer can even in high quality mimic an analogue equalizer in the digital domain, for theoretical reasons. people will speak about "frequency transforms" and stuff as if they know what they're talking about (undergrad EE courses), but there are boundary conditions to every normally known digital filter that make it impossible to completely correct. Not by a void in the vertical bit resolution, but because filters require knowledge of the signal between the samples (in time sense), and the only correct theoretical fashion in which to compute the signal between the samples takes an infinite integral.

 

So it's hopeless then, those cold and harsh digital signals are all we're gonna get ? No, there are tape and electronic and (electro-)mechanical instruments! And since very long ago intelligent and informed people have worked on preventing the errors digital causes, and prepared mixes for studio processing, and built in DAC preparation and correctable filters in certain digital instruments and tools.

 

Theo V.

[Mike Rivers]

 

But if one of the inventors on the CD team publishes that the bit rate was too slow and caused severe quantization errors, I figure he knows a lot more about the subject than I do.

 

He may know more than you do, but he didn't express it very well. Quantization error isn't a result of too low a sample rate, it's a result of the A/D converter not being able to accurately measure the voltage at the time the sample is taken. Proper dithering can eliminate quantization errors, with the tradeoff being more noise. It may be that this inventor, for simplicity, lumped all differences between input and output as "quantization" since that's what the overall process is.

 

But I repeat, if the general public really cared about audio quality, the SACD wouldn't have died wasted death.

 

Nor would Pono, and high resolution downloads and streaming, while still alive, aren't exactly doing a booming business in high resolution audio. When it comes to music, people value convenience over quality, mostly because they are usually doing something else when listening to music. There's sufficient high resolution material available to keep those with the $1,000 DACs and $2,000 headphones happy when they're relaxing in the easy chair with their favorite symphony or pop star.

 

It seems the public likes video resolution more than audio judging from the screens I see in the stores.

 

There is really some pretty amazing video out there, but people enjoy different things about video than they do about audio. Video scenes are short and are always changing, and that's part of what keeps watchers interested, whether it's a football game, an undersea adventure, or a couple of porn stars going at it. It's pretty amazing to see a close-up of a shark's wide open mouth for a few seconds, but you probably wouldn't enjoy hearing a song where someone was pushing Solo buttons on the console throughout the piece so you could hear a couple of seconds of pick noise during a guitar solo or a backup singer taking a breath before coming in on the chorus.

 

 

[Mike Rivers]

 

You're probably thinking of E-Mu's Emulator II. I did the manual for it and amazingly, still have it but haven't booted it up in a while. They actually got 12-bit operation through companding.

 

That might be it, or maybe it was the Ensoniq EPS (thanks, Google) that I was thinking of. Check the battery in your E-II. It might be leaking.

 

As I recall, E-Mu made some of the better quality PC sound cards, too. And they also had a hard disk 8-track recorder, the Darwin, for a short time.

 

[Mike Rivers]

The problem is that there's a mathematical "reconstruction theorem" that states you need a near infinite resampling/oversampling filter in order to get close to retrieving the samples that you've made as in the original analogue sound quality.

 

The mathematical theorem is that you need only two samples per cycle of the highest input frequency that you want to digitize in order to perfectly reproduce the sound at the output. It's not just a good theory, IT'S THE LAW. But you have to obey the law if you want to be perfect (according to the law, that is) and that's difficult to do. You must set your boundaries and stay within them. If we accept that there's nothing audible above 20 kHz, a 44.1 kHz sample rate is perfectly adequate. You can have a 1-bit sampler and still reproduce the frequency accurately. More bits means less noise along with the reproduction, which is why we use more bits.

 

How do we assure that nothing above 20 kHz gets to the A/D converter? By putting a low pass filter ahead of it. But a filter built from real world capacitors, resistors, and inductors doesn't cut off sharply at a specified frequency, and every little squeak above the Nyquest frequency (2x the sample rate) will come back as some other frequency. In the early days, they approached the problem the only way they knew how - by making a multi-stage 20 kHz filter that started going to work as low as 1 kHz. It's not like the frequency response would start dropping off above 1 kHz, but rather, that different frequencies were delayed by different amounts of time (microseconds or milliseconds, not slapback echos) and that caused things to sound different coming out than when they went in.

 

Today's A/D converters "oversample," making the effective sample rate several times higher than the highest frequency you care about, and use digital filtering to construct a filter with minimum phase shift ("group delay" is the correct term). But there's always a compromise, and in this case it's some roughness in the filter's passband and, because digital filtering uses a string of delays, an effect called "pre-ringing" that can be audible in a filter that's not well designed.

 

So, nothing's perfect, and it never will, be. But then neither is analog tape. It's a matter of what you least like to hear that determines your preference. In the 1980s, there were good reasons why digital audio didn't sound good. That today's digital audio can't sound good is no longer a valid technical argument.

 

 

[Theo Verelst]

Hey, I'm a theoretically well schooled EE, I know what I'm talking about.

[Mike Rivers]

Me, too. Sampling is about information theory, not exclusively Electrical or Electronic Engineering.

[Theo Verelst]

You're a decent or better university educated EE ?

[Notes_Norton]

 

But if one of the inventors on the CD team publishes that the bit rate was too slow and caused severe quantization errors, I figure he knows a lot more about the subject than I do.

 

He may know more than you do, but he didn't express it very well. Quantization error isn't a result of too low a sample rate, it's a result of the A/D converter not being able to accurately measure the voltage at the time the sample is taken. Proper dithering can eliminate quantization errors, with the tradeoff being more noise. It may be that this inventor, for simplicity, lumped all differences between input and output as "quantization" since that's what the overall process is.

 

In all fairness, that could be me mis-remembering the why of what I read in an article on the 25th anniversary of the CD. He definitely said severe quantization errors, and there is a great possibility you are more correct on the why and I just mixed my facts up.

 

I have an AA degree in electronics, but none of it was digital. There was the computer option and the communications option (mostly RF). At that time computers took up an entire building and there were few jobs in the computer field, so I was advised to take the communications road.

 

But at the same time I was first tenor sax in the all-state band every year I could participate, and section leader which goes to the first alto by default. Music seemed like a lot more emotionally rewarding way to make a living, and I don't regret making that choice for one minute.

 

Notes

 

[Anderton]

That might be it, or maybe it was the Ensoniq EPS (thanks, Google) that I was thinking of. Check the battery in your E-II. It might be leaking.

 

As I recall, E-Mu made some of the better quality PC sound cards, too. And they also had a hard disk 8-track recorder, the Darwin, for a short time.

 

The order of appearance was Emulator (1981), Emulator II (1984), Ensoniq Mirage (1984), and EPS (1988). The Emulator II got the glory because it did what previously was doable only with Fairlights and Synclaviers, but at a much lower price point. The Mirage was far more affordable, around $1,700, but the sound quality and OS was sketchy compared to the EII.

 

Does anyone even make serious hardware samplers any more, not just phrase samplers and such?

[Mike Rivers]

Mirage! That's the name I was looking for. 8 bits, about $1200.

 

Does anyone even make serious hardware samplers any more, not just phrase samplers and such?

 

Well, the Fairlight is now owned by Blackmagic, but what I saw of it at InfoComm looked like it was dedicated to video production - not surprising because that's what Blackmagic is about. It's part of a larger software/hardware system, not really designed for playing music, but apparently video editors like it for audio-for-video.

[Notes_Norton]

I still have an Akai S900, but I don't use it anymore. If there was a market for it, I'd sell it. But since it has been probably 20 years since I powered it up, it would have to be as-is. The floppy disk drive is mechanical and hasn't spun in all those years. Even if I powered it up and it worked, I don't know if it would work for long.

 

It was a good tool in it's day. I sampled my J-Bass, a right and left hand snare drum (so the rolls don't sound like machine guns), some deadened kick drums, and some exotic percussion. Anything else I sampled has been long forgotten by improvements in synthesizers. I still use the bass and snares at times, playing them on an archaic Peavey sample player (anyone remember those?)

 

Notes

 

[Anderton]

I still use the bass and snares at times, playing them on an archaic Peavey sample player (anyone remember those?)

Yes, I have a Peavey SP! Wrote the manual for it. Not sure it still works, a lot of old Peavey gear was subject to the Batteries of Death.

 

 

[zeronyne]

Does anyone even make serious hardware samplers any more, not just phrase samplers and such?

 

Yes, but they are baked into workstations. I have an Akai S5000, and in terms of audio processing, the Kronos does a fine job keeping up.

 

And Craig, I think he's talking about the Mirage (the EPS, like the EII, was 12 bit). I would not have called the EII "affordable" to a working musician.

 

[Anderton]

And Craig, I think he's talking about the Mirage (the EPS, like the EII, was 12 bit). I would not have called the EII "affordable" to a working musician.

 

Probably not But compared to a Fairlight or Synclavier, it really was a breakthrough.

 

[Notes_Norton]

Yes, I have a Peavey SP! Wrote the manual for it. Not sure it still works, a lot of old Peavey gear was subject to the Batteries of Death.

 

 

I have two, it's been so long, I don't know if the spare works.

 

It's a nice piece of hardware, although archaic by today's standards. Fitted with flash memory instead of that turtle-speed floppy disk and it would still be a one-rack-space piece of useful gigging gear.

 

Notes