Digitally Enhanced Vinyl Poll

[dallasjustice]

Michael, in the context of this thread...may one substitute 'analog' for 'vinyl'?

No, you may not.

I really love your question though. My theory is that some folks want to kick vinyl to the curb in favor of digital. On paper, this makes a lot of sense. Digital has some nice specs over vinyl. BUT I don't think that the reason so many folks still prefer vinyl has been fully investigated or well understood. When people say they prefer vinyl over digital, I take them at their word. But I also wonder why. I think there's some type of "distortion" inherent in vinyl (maybe other analog sources too) which could account for this preference. Even though I'm a hardcore digital geek, I think it's throwing the baby out with the bath water for me not to give vinyl a chance (in my own way).

 

[Bruce B]

I don't really follow. As far as I know, the RIAA curve is simply an EQ applied to the viny signal to compensate for the lack of linearity inherent in the format. It's a fairly gentle curve so I don't really understand why you think 64 bit floating point FIR filters could easily do the RIAA compensation.

An RIAA curve is not "gentle" at all. You know what would happen if you tried to create digital filters to boost/cut this much?? How many filters would you need? I've heard experts talk about this. I'll have to search for my emails and such on what I received. Holy crap... boosting and cutting 40+ at the extremes??? From mV to line level??

 

[Bruce B]

In reference to a Stereophile article, Ripping LP's with Pure Vinyl, we had a discussion with an electronics digital guru in our club and his statement was:

In response to the "ignorance is bliss" camp and the "hit it with a hammer until it works" crowd I wish to comment on the below Stereophile event before others take it as absolute truth in advertising and waste a lot of time making bad rips of their LP collection.

Bruce's comment is appropo, and while approximately duplicating the RIAA curve in the digital domain is possible using a big enough hammer, ie., 64 bit floating point arithmetic, two resistors and two capacitors will do the job perfectly, no sweat involved. That is because those four components perfectly undo the reciprocal action of the RIAA preemphasis which has been used to cut the record.

Now, on to the real fallacy of plugging your cartridge into a professional A-D converter designed to handle +24dBu input signals (that's 12.28 volts rms guys.) S/N and Zin! (Signal-to-noise ratio and input impedance.) The typical input impedance of this breed of converters is between 10K and 20K ohms for a line level input or, if it has mic preamps built in, 2-5K ohms - do either of these look like the impedance your pickup was designed to work into? Moving magnet and moving iron pickups need not apply as they like 47K ohms and a little bit of capacitance, something not specified in most pro gear because they assume your source is low impedance and doesn't care about 'C'.

Ok, maybe you have the mic pre version of converter and your $5K moving coil pickup is OK with 2-5K ohms loading, and the mic preamp in question has an astoundingly low input noise spec of -132dBu (this is usually only found in stand-alone preamps costing upwards of $3-5K designed to work with ribbon microphones,) then things will be well if the A-D has at least 120dB signal to noise at full scale and you've adjusted levels to use that entire dynamic range. Then the minute signal from your vinyl will sail through to the dsp engine and....well just whip out MATHCAD and crank away at that curve. This is where more processing power may pull you through and probably do as well as a decent phono preamp coupled to a half decent A-D.

I rest my case.

 

[amirm]

Funny coincident Bruce as I was reading the Pure Vinyl web site as you post that: http://www.channld.com/pure-vinyl_faq.html

Have you listened to it?

 

[Bruce B]

Funny coincident Bruce as I was reading the Pure Vinyl web site as you post that: http://www.channld.com/pure-vinyl_faq.html

Have you listened to it?

Oh yeah. It's fine if you can't afford a decent system. Otherwise........

 

[dallasjustice]

Thanks Bruce. That is a fairly gentle slope. With digital filters, it's the steepness of the slope that could cause problems like preringing. In this case, the slope is very gradual. Sure, there's a large difference between the low end and high end, but the slope is very smooth. This wouldn't be difficult at all to achieve without audible digital artifacts.

I asked Bernt to post a comment here. Maybe we could hear from someone who knows a little about how FIR filters work. I don't pretend to be any kind of "expert." But Ive worked with these types of filters before so I know a little about their behavior.

An RIAA curve is not "gentle" at all. You know what would happen if you tried to create digital filters to boost/cut this much?? How many filters would you need? I've heard experts talk about this. I'll have to search for my emails and such on what I received. Holy crap... boosting and cutting 40+ at the extremes??? From mV to line level??



View attachment 21308

 

[asiufy]

Michael,

That sounds like a fun exercise
While you're at it, why don't you go all-out, and try doing the digitalization post-RIAA, with a proper phono stage, and report the results? You can use something like this:

http://www.psaudio.com/products/nuwave-phono-converter/

Or if you want really cheap, this:

http://adl-us.com/product/gt40-usb-dac/


alexandre

 

[dallasjustice]

When it comes to A to D/D to A, the Lynx Hilo is one of the best. It even edged out the merging Hapi in this loopback test.
https://www.gearslutz.com/board/11083311-post1131.html

PSaudio isn't something I'd be intereted in. Their best DAC has awful measurements even after their software update.

Michael,

That sounds like a fun exercise
While you're at it, why don't you go all-out, and try doing the digitalization post-RIAA, with a proper phono stage, and report the results? You can use something like this:

http://www.psaudio.com/products/nuwave-phono-converter/

Or if you want really cheap, this:

http://adl-us.com/product/gt40-usb-dac/


alexandre

 

[CGabriel]

Okay. I understand that I haven't made myself very clear. I appreciate Bruce and Caelin's perspective. However, I want to use digital to enable all of the DSP I already do. I don't plan on ripping vinyl. I will setup a "live" vinyl rig which gets converted to digital so that I can apply digital crossovers and target curve in Audiolense as well as an "enhanced" RIAA curve.

I want to contemporaneously get all the benefits of vinyl and all the benefits of digital.

OK, I see that makes sense. If you can gain the ability to correct what can be obvious and significant acoustic and speaker related problems and you only give up some slight loss of fidelity to the vinyl source - the pros could easily outweigh the cons.

Personally I use digital correction in the digital music chain and prefer to bypass it when using the turntable but that is just a preference in my own room.

 

[CGabriel]

In reference to a Stereophile article, Ripping LP's with Pure Vinyl, we had a discussion with an electronics digital guru in our club and his statement was:

In response to the "ignorance is bliss" camp and the "hit it with a hammer until it works" crowd I wish to comment on the below Stereophile event before others take it as absolute truth in advertising and waste a lot of time making bad rips of their LP collection.

Bruce's comment is appropo, and while approximately duplicating the RIAA curve in the digital domain is possible using a big enough hammer, ie., 64 bit floating point arithmetic, two resistors and two capacitors will do the job perfectly, no sweat involved. That is because those four components perfectly undo the reciprocal action of the RIAA preemphasis which has been used to cut the record.

Now, on to the real fallacy of plugging your cartridge into a professional A-D converter designed to handle +24dBu input signals (that's 12.28 volts rms guys.) S/N and Zin! (Signal-to-noise ratio and input impedance.) The typical input impedance of this breed of converters is between 10K and 20K ohms for a line level input or, if it has mic preamps built in, 2-5K ohms - do either of these look like the impedance your pickup was designed to work into? Moving magnet and moving iron pickups need not apply as they like 47K ohms and a little bit of capacitance, something not specified in most pro gear because they assume your source is low impedance and doesn't care about 'C'.

Ok, maybe you have the mic pre version of converter and your $5K moving coil pickup is OK with 2-5K ohms loading, and the mic preamp in question has an astoundingly low input noise spec of -132dBu (this is usually only found in stand-alone preamps costing upwards of $3-5K designed to work with ribbon microphones,) then things will be well if the A-D has at least 120dB signal to noise at full scale and you've adjusted levels to use that entire dynamic range. Then the minute signal from your vinyl will sail through to the dsp engine and....well just whip out MATHCAD and crank away at that curve. This is where more processing power may pull you through and probably do as well as a decent phono preamp coupled to a half decent A-D.

I rest my case.

Thanks Bruce. I don't understand why we are talking about doing the RIAA in the digital domain. When we do a vinyl to DSD rip we use a good phono preamp before the AD conversion. To get even remotely good results the analog front-end needs to be as good as you can get including the phono pre. RIAA has been around for some time and we have gotten pretty good at using it in the analog domain.

 

[Bruce B]

Thanks Bruce. I don't understand why we are talking about doing the RIAA in the digital domain. When we do a vinyl to DSD rip we use a good phono preamp before the AD conversion. To get even remotely good results the analog front-end needs to be as good as you can get including the phono pre. RIAA has been around for some time and we have gotten pretty good at using it in the analog domain.

That's what needs to be done. This is how I do it as well. Don't know where Michael is going with this....

 

[dallasjustice]

That's what needs to be done. This is how I do it as well. Don't know where Michael is going with this....

Bruce,
See the infra yellow line SPL response at 1/12th per octave smoothing. That kind of seated position linearity sounds great. IME, it's simply not possible to achieve that kind of full range linearity without strategically placed subs, digital crossovers, proper delay and target curve using perceptual windowing. This is what I call high resolution because these measurements correlate very nicely with better music playback. My objective is to achieve this type of linearity with a turntable as a source without analog EQ. IOW, a purely analog system can't achieve this type of linearity at seated position where my ears hear the music. To my knowledge, nobody here has ever done anything like that before and I think it would be very fun to try it out.

 

[Bruce B]

When it comes to A to D/D to A, the Lynx Hilo is one of the best. It even edged out the merging Hapi in this loopback test. .

Really? Besides... I have a Horus.


Lynx Hilo (CoolColJ) - 6 last second part disregarded
-0,1 dB (L), -0,1 dB (R) Corr Depth: 41,8 dB (L), 43,7 dB (R), Difference*: -59.1 dBFS (L), -59.2 dBFS (R)

Merging Technologies Hapi
ADA8 ---> ADA8 (ljudatervinning): 0.391dB (L), 0.376dB (R) Corr Depth: 38,5 dB (L), 40,1 dB (R) Difference*: -58,9 dBFS (L), -60,3 dBFS (R)

MSB Platinum IV Plus DAC ---> MSB Studio ADC master (DUJS)
4,9 dB (L), 4,9 dB (R) Corr Depth: 62,8 dB (L), 63,4 dB (R) Difference: -77,7 dBFS (L), -77,4 dBFS (R)

Sorry Michael... .I don't listen to "loopback test"... I only listen to music!

 

[juicehifi]

Thanks Bruce. That is a fairly gentle slope. With digital filters, it's the steepness of the slope that could cause problems like preringing. In this case, the slope is very gradual. Sure, there's a large difference between the low end and high end, but the slope is very smooth. This wouldn't be difficult at all to achieve without audible digital artifacts.

I asked Bernt to post a comment here. Maybe we could hear from someone who knows a little about how FIR filters work. I don't pretend to be any kind of "expert." But Ive worked with these types of filters before so I know a little about their behavior.

Bernt here,

Like everybody else I played a lot of vinyl when I grew up. I put it away when the CD became the dominating medium, so my experience with high end vinyl systems is very limited. But I do like the "less is more" approach to mixing and mastering - and that seems to be the norm more so with vinyl than with digital media these days, so I would not be surprised if there is a lot of music out there that sounds better on vinyl.

RIAA was implemented in Audiolense to support users who wish to combine digital sound correction with vinyl playback. They have to get the signal into the digital domain anyway, and that is by far the best place for any sorts of processing - at least if the signal is present in the digital domain to begin with. The standard RIAA and the enhanced RIAA has been implemented. They were chosen because they appeared to be the two that were both neutral and relevant. Further voicing can be done the regular way in Audiolense if needed. The time coefficients used are the same as the official ones. And in any case, the user can examine the frequency, phase and time domain behavior of both RIAA filters in Audiolense. The frequency response is calculated by using straight forward math with double precision numbers. This frequency response is turned into a minimum phase impulse response with the same frequency response. This RIAA filter can be run as a stand-alone filter or merged with a correction filter.

The frequency and time domain behavior of an RIAA should be the same whether it has been implemented by a passive network, an active network, digital biquads or a FIR filter - as long as the spec has been followed. In practice it is numerical rounding errors in the digital domain vs various degrees of distortion in the analog domain. The rounding errors for single and double precision floating point math will typically be way below what any active hifi component can discriminate and for all practical purposes negligible. I am aware that there are a few analog RIAA solutions out there who has (more or less intended) deviations from the spec, and one should expect them to sound tonally different from those who are made to spec.

The 40 dB attenuation of the low frequences on the vinyl is very easy to equalize by FIR filters. But the signal that enters the pickup appears to be fragile, and same care as always should be taken to preserve the signal integrity as well as possible and amplify the signal to an appropriate line level before the AD conversion. In conjunction with the to the two RIAA filters in Audiolense there is an amplification /attenuation function that enables the user to dial in the right level overall after DA conversion and RIAA correction.

The RIAA in Audiolense has not been tested in "public" to my knowledge so I am curious to see how this works out in practice. In the mean time I'll be happy to answer any further questions about the Riaa correction and Audiolense in general.

 

[amirm]

Hi Brent. Welcome to the forum! Much appreciate the explanation. Can you give us some speeds and feeds? What sampling rate does the RIAA eq run at? Is it at whatever rate the data was digitized or does it downsample? What is the CPU usage at the highest sampling rate?

 

[bonzo75]

Surprised Marty hasn't joined, because in his system, the Goldmund Studio (not reference) going through the TacT slammed the Meitner. I also heard another Linn top of the line TT going through a Trinnov sound much better than the Linn Streamer going through the Trinnov, on large Adam active speakers (Tensor Alpha)

 

[CGabriel]

I think that Bernt mentioned something very key in passing which is the signal level before AD conversion. This may be highly hardware dependent and the method (analog amplification stage) may be critical to preserving the signal.

 

[dallasjustice]

Thanks Bernt for posting here!

IIRC, audiolense RIAA settings include two RIAA filters as well as impedance mismatch compensation. Is that correct?

I finally procured a TT and I think it's got a MC cartridge. This means there might be some HF rolloff due to my mic pre's lower input impedance, right? Is there a compensation filter for this or would I just need to manually adjust the target curve in the HF?

Bernt here,

Like everybody else I played a lot of vinyl when I grew up. I put it away when the CD became the dominating medium, so my experience with high end vinyl systems is very limited. But I do like the "less is more" approach to mixing and mastering - and that seems to be the norm more so with vinyl than with digital media these days, so I would not be surprised if there is a lot of music out there that sounds better on vinyl.

RIAA was implemented in Audiolense to support users who wish to combine digital sound correction with vinyl playback. They have to get the signal into the digital domain anyway, and that is by far the best place for any sorts of processing - at least if the signal is present in the digital domain to begin with. The standard RIAA and the enhanced RIAA has been implemented. They were chosen because they appeared to be the two that were both neutral and relevant. Further voicing can be done the regular way in Audiolense if needed. The time coefficients used are the same as the official ones. And in any case, the user can examine the frequency, phase and time domain behavior of both RIAA filters in Audiolense. The frequency response is calculated by using straight forward math with double precision numbers. This frequency response is turned into a minimum phase impulse response with the same frequency response. This RIAA filter can be run as a stand-alone filter or merged with a correction filter.

The frequency and time domain behavior of an RIAA should be the same whether it has been implemented by a passive network, an active network, digital biquads or a FIR filter - as long as the spec has been followed. In practice it is numerical rounding errors in the digital domain vs various degrees of distortion in the analog domain. The rounding errors for single and double precision floating point math will typically be way below what any active hifi component can discriminate and for all practical purposes negligible. I am aware that there are a few analog RIAA solutions out there who has (more or less intended) deviations from the spec, and one should expect them to sound tonally different from those who are made to spec.

The 40 dB attenuation of the low frequences on the vinyl is very easy to equalize by FIR filters. But the signal that enters the pickup appears to be fragile, and same care as always should be taken to preserve the signal integrity as well as possible and amplify the signal to an appropriate line level before the AD conversion. In conjunction with the to the two RIAA filters in Audiolense there is an amplification /attenuation function that enables the user to dial in the right level overall after DA conversion and RIAA correction.

The RIAA in Audiolense has not been tested in "public" to my knowledge so I am curious to see how this works out in practice. In the mean time I'll be happy to answer any further questions about the Riaa correction and Audiolense in general.

 

[Brucemck2]

Many years ago I ran my analog rig (Basis vacuum, Vibraplane, Graham + Koetesu, Aesthetix Signature) through a heavily modified DEQX. I did that for the simple reason that at the time my line arrays lacked passive crossovers and required the DEQX.

The DEQX provided preamp functionality, three way crossover functionality, and driver time alignment (two way line arrays crossed to Seaton subs, with filters set up by Mark Seaton) while a downstream QSC provided PEQ for the Submersives. Even with all the dsp the setup provided a really nice "analog feel" that differed from purely digital feeds. A good fraction of that was the inherent Koetsu signature and rich tube sonics of the Aesthetix shining through.

 

[dallasjustice]

The DEQX is very transparent. That sounds like a really cool system.

I know you use Acourate. Have you thought about setting it up again using Acourate or Audiolense?

Many years ago I ran my analog rig (Basis vacuum, Vibraplane, Graham + Koetesu, Aesthetix Signature) through a heavily modified DEQX. I did that for the simple reason that at the time my line arrays lacked passive crossovers and required the DEQX.

The DEQX provided preamp functionality, three way crossover functionality, and driver time alignment (two way line arrays crossed to Seaton subs, with filters set up by Mark Seaton) while a downstream QSC provided PEQ for the Submersives. Even with all the dsp the setup provided a really nice "analog feel" that differed from purely digital feeds. A good fraction of that was the inherent Koetsu signature and rich tube sonics of the Aesthetix shining through.