LP with better dynamic range than digital
- Thread starter amirm
- Start date
Yes you can record and listen to a tone recorded at lower than -96 db with dithered 16 bit. You will have to boost the level up until the tone and the noise are audible. While it is available we don't need mathematical proof as we can construct a signal and hear it ourselves.
I think what is confounding about all this is the generic use of noise. Noise can be white, pink, shaped or other shapes. How do you measure it? You specify an RMS level over a given bandwidth. I actually constructed a shaped noise through which I was able to hear a tone 50 db below the level of noise over the 20-20khz band. That was by having the noise low in the region of the tone and much higher elsewhere. You will see academic research when referring to white or pink noise say humans can hear 15 to 20 db into noise. With other shapes of noise a bit more is possible. So just using the term noise can be confounding if it isn't specified in this context of digital audio.
I imagine John at Benchmark had in mind -140 db for properly dithered tone in the 3-4 khz region with shaped dither. You will have to amp that signal of course, but yes it is just possible to perhaps manage that. I managed to hear -130 db on a 3 khz tone using headphones with a shaped dither in 16 bit file. I could hear -120 db with triangular (or near flat dither). So that dithered 16 bit is capable of portraying lower than -96 db and his claim of near equivalent to -120 db range isn't wildly wrong. I wouldn't quite characterize it that way. But then every time the subject comes up it seems confusing to people.
By the way your complaint about UV22 pushing noise into the 18-22 khz region being a problem also doesn't quite matter. Our thresholds for hearing those tones is very high in the first place. If you put noise there, it can still be below the point at which our hears would hear a tone even without noise. Which allows a lower noise floor in the frequencies at which we can hear with a low threshold. So human hearing actually doesn't have a dynamic range that is even over frequency. That is why shaped noise is audibly better than a flat noise spectra.
I can show you FFTs where the tones are visible even though lower than the RMS 20-20khz value of the noise. I can construct a sample file and send it to you. But it sure is a shame this kind of confusion happens many thousands of times over and over to only have someone anew make the same mistaken complaint saying it just can't be so.
The same conditions apply to LP noise. But the LP noise is at a higher level upon playback. LP simply doesn't have the dynamic range encoding ability of CD. It simply doesn't. Recordings may or may not have more on one or the other, but the CD has more dynamic range to work with. With dither something more than 96 db.
Although you need to state this as noise shaped dither and then one needs to consider its implications that has been touched upon in the past in the engineering world by such as Bruno Putzeys; it is not a free lunch and can have further implications further down the processing-function chain as well.
Cheers
Orb
Well no, just referring to the quote of me in your post. You can do this with flat dither and recover audible tones below -96 db. You can recover them lower if you shape the dither. Yes of course if you add noise and then process it with EQ or compression or other editing tricks downstream it can become audible. But the same would be true of post processing of LP noise or tape noise.
And you normally do not use shaped dither with PCM ADC audio or mixing (in context of this thread).
I am thinking more from ADC side of this debate and native 16bit PCM.
Cheers
Orb
Last edited:
We did cover this earlier in the thread- in that it was agreed that 10db into the noise floor is possible. Given that it was not that hard to get to 100db dynamic range on an LP. Apparently, that 10db figure (or 20, or 30, depending on how much kool aid was drunk) won't work unless the noise floor truly is noise like white noise. If it is composed of harmonic and inharmonic distortions, the ear apparently cannot penetrate the noise floor.
The bottom line though apparently remains the same and and functions in the face of the limits of the technology: LPs will continue to have greater dynamic range in practice due to the practices of the recording industry.
Tomelex, I recommend you spend some time around an LP mastering machine, and compare your experiences with the same recording held in the digital domain. You might want to rethink your assertions as seen in post 237. In a nutshell I get the impression that you've not spent much time with state of the art LP playback. Another way to look at this: If it is distortion that people prefer about the LP, what does that say about human hearing? I am going to flip that around for a moment, maybe the answer will become clearer: if people don't like the small distortions of digital, what does that say about human hearing?
The answer is of course that some distortions are much more objectionable to the ear than others, even though our instruments record them at lower levels.
If we really want to create state of the art, IMO/IME we need to know how the human hearing/perceptual rules work, such that we make technology that conforms more closely to the things that the ear cares about rather than those things that look convenient and nice on paper. In this regard I am not saying LP is better ultimately (although I am saying it is better now); I am saying that if you want **any** recorded media to really shine, it has to pay attention to how our ear/brain system really works.
I think the answer is digital systems have distortions below the audible level of humans. It therefore can transparently convey information for listening. From that follows the idea if a large percentage of people have a preference for a medium with higher and audible distortions it isn't a failing of the digital audio system for lacking those distortions. If one wishes to cater to that preference, one can process digital audio. Digital audio isn't insufficient for the purpose in it basic design.
Now there is also some knowledge as to why LP distortions might be preferred. More research into exactly what part of it is preferable would help. Some fair amount is already known. Sensory stochastic resonance is not too dissimilar to dither in digital systems. It is a process where the right type of noise added to a signal can allow detection of a low level signal that otherwise is not detected. As psycho-acoustics show the ear to have a short term SNR of about 60 db the LP which in most systems has noise levels around that 60-70 db mark may just happen to be at the right level. Hence enhanced low level detail. The answer wouldn't be to do anything more than add such noise to a digital recording. The perceived better dynamics in LP is also understand at least somewhat. Most cartridges on higher level material on the peaks distorts by several percent. It is known in psycho-acoustics that such signals increase in apparent loudness. What happens is a signal can be heard at a given level and another nearby will need to be much louder, but one further away in frequency activates additional bands the ear uses. Extra bands being activated will increase subjectively perceived loudness substantially when the total RMS level is actually a quite small increase. So you hit a peak with LP, distortion adds audible harmonics in several bands that prior to the peak were inaudible and there is your big dynamics. Same signal in digital is clean, undistorted and sounds more restrained. One could have DSP to process peaks for the proper distortion spectra to be added.
Now if the approach taken is digital is broken or inferior you aren't getting anywhere because you are trying to fix what isn't broken. If the approach is we prefer LP type analog sound, then researching what DSP can give us that sound has the chance to be successful.
And just a note, I don't agree with, "Given that it was not that hard to get to 100db dynamic range on an LP."
Partially I agree, but not when it comes to classical music on digital. There engineers usually try to take advantage of available dynamic range (a few exceptions here and there notwithstanding).
Indeed. This bears highlighting rather than being buried in the middle of a post.
Edit: Beaten to it by Al. I should qualify by saying that my own musical tastes do not normally include classical or other genres that routinely employ dynamic range as a musical effect.
Al. M,
Classical music is a too broad genre to support such claim. Most of my chamber music LPs sound more dynamic than the CD versions.
You might be misinterpreting the comment. I was simply adding up numbers (which was 'not that hard'), some of them presented by others here; at any rate actually getting 100 db on an actual LP in practice would be pretty challenging. You would have to have all the stars perfectly aligned to get the noise floor right.
The only way I know to do it would be direct-to-disc, but the noise floor of the room itself would probably mean that such a thing is practically impossible so it would have to be an electronic music recording. Same for digital as far as that goes.
Okay, misunderstood you. One wonders if a 100 db capable LP might be declared dead and lifeless like digital were it done.
I completely agree with Atmasphere that "If we really want to create state of the art, IMO/IME we need to know how the human hearing/perceptual rules work, such that we make technology that conforms more closely to the things that the ear cares about rather than those things that look convenient and nice on paper."
I also agree with esldude that digital is most likely the way that this will be achieved in the future but I don't agree that digital, at the moment, is already at a stage where it is a suitable platform for achieving this.
My sense, at the moment, is that the noise profile of both LP & digital needs to be studied in detail & this might reveal why we seem to perceive better dynamics with LP than with most forms of digital. It has already been mentioned that we hear below the noise floor as long as this noise floor is white noise - modulation of this noise might affect our perception of dynamics & some modulation might be more disruptive than others?
I would like to see some investigation/characterisation of the noise modulation aspects of S-D DACs Vs true multibit DACs Vs LP playback as a first approach. I suspect it may reveal some clues about our perception. Some analysis & measurements of DACS was done by Stereophile up to 1997 (based on a proposed a new technique by Dr. Richard Cabot for measuring noise modulation in D/A converters) but they didn't seem to find correlation with listening impressions
I also agree with esldude that digital is most likely the way that this will be achieved in the future but I don't agree that digital, at the moment, is already at a stage where it is a suitable platform for achieving this.
My sense, at the moment, is that the noise profile of both LP & digital needs to be studied in detail & this might reveal why we seem to perceive better dynamics with LP than with most forms of digital. It has already been mentioned that we hear below the noise floor as long as this noise floor is white noise - modulation of this noise might affect our perception of dynamics & some modulation might be more disruptive than others?
I would like to see some investigation/characterisation of the noise modulation aspects of S-D DACs Vs true multibit DACs Vs LP playback as a first approach. I suspect it may reveal some clues about our perception. Some analysis & measurements of DACS was done by Stereophile up to 1997 (based on a proposed a new technique by Dr. Richard Cabot for measuring noise modulation in D/A converters) but they didn't seem to find correlation with listening impressions
Hmm, I hear you, but I also sense quite a bit of sugar-coating. For example, how can you hear a -130dB 3kHz tone with 16-bit shaped dither if 16-bit cannot even encode such a signal, or to encode it as you say, it has to be amplified. But let's leave semantics aside: can anyone mathematically prove digital's alleged superiority when it comes to dynamic range.
I appreciate the sound of vinyl, but if we measure real world, real consumer playback I would hope you agree vinyl has to take a back seat as far as useable real world dynamic range. the topic of this post.
Now, while I do not have the best playback gear (and who does?) I can record an LP to my digital 24/96 recorder (which records and plays back at 24/96 no conversions) and play it back and the playback sounds the same to me. It captures the things vinyl does, it does not hurt then in any way I can hear.
Again though, the OP is not sound, but spec. LP does pretty good for itself and it is a fully mature technology, and not likely to advance as far as at the cutter end IMO.
Now, if digital (not just redbook, but digital) is such a more measureably accurate reproducer of sine tones, and if I record LP to it and play it back it sounds near enough identical for me, but, folks who master it master it for cars, then that is not digital fault. And yes, digital is what now, 30 some years old in the consumer market, and has been steadily improved over time.
Its like tape, digital far exceeds tape, but folks like that tape effect, odd ordered harmonics and extra splash on peaks and less separation and etc, just like the less steller specs of consumer vinyl sound pretty darn good.
Do you use a digital delay in your chain? Just funin Atmasphere.
OK, so what were we talking about.....ahahahahahha
In my keyboard setup, I use digital delay all the time. I just can't get the spring reverb to do what I want on stage, and it can go into feedback like a microphone.
Bold: tape has a lot of separation! - enough that that is by no means one of its weaknesses.
Regarding advances in cutter technology, I can name one... the cutter only uses a fraction of the power of the cutter amps. This is done so that the amps can never be overloaded. The problem is that the cutter uses so little power that its actually operating most of the time at a power level below which the amplifier's distortion actually increases rather than decreases. This is due to phase splitter issues and the application of loop negative feedback in the cutter amplifier itself, and can be observed on most push-pull amplifiers (including class D) made, whether tube or solid state.
We solved that, by using an amplifier whose distortion level falls linearly to unmeasurable as power is decreased. The amp does that by not running feedback and not having a dedicated phase splitter, despite being push-pull. It gets its low distortion out of inherent linearity, but still has the advantage of a 30db feedback loop provided by the cutter electronics for anti-resonance. The result of this is greater low level detail and its easily heard as there is less masking.
However my main reason for responding to this post is the opening comment in the quote above. 'As far as useable real world dynamic range. the topic of this post.' is concerned, the LP will continue to have it all over digital formats for the foreseeable future for no other reason than I have been harping on for a number of pages now, which is in the 'real world' the record industry is going to compress the hell out of the digital edition of the album, because they can and expect to play it in a car. That is the 'real world'; we audiophiles live in a sort of fantasy place where we try to push the limits, and the record industry could care less what we think. But they know that an LP won't be played in a car, so they don't compress it nearly so much.
I would appreciate it very much if you would just go out and change that for us
I know I can't do it... I would love it if my CDs (or whatever) got the same impact as the same recording on LP. That impact (especially in the bass) is one of the reasons I hold out for the LP if its available.I think this is where the problem might lie - are reproducing sine tones a good way to determine how well it will behave with real-world music? As Opus11 used to say, it's about the crest factor of music - "my gut feel is we need a multitone test waveform rather than a single sinewave as a sinewave has a crest factor quite unlike music and if a sine provoked the condition we'd see it on the 'N' part of a THD+N vs level test."
So really, it's academic to talk about measured DNR, if the perceived dynamic range (or shock factor) doesn't concur with the measurement. There are also the signs of some anomalies in these measurements, which indicate noise modulation - see this post
"The dynamic range is impressive at 130dB but rather unfortunately the dynamic range is not maintained when the signal level rises above -40dB. Eyeballing the line shows a clear kink in the otherwise perfect trajectory. In going from -40dB to -30dB the signal goes up 10dB but residual D+N, which previously held constant climbs 6dB. This is classic noise modulation as the FFT plot shows the residual to be very unlikely to be distortion - that's going to be a much lower level. Noise modulation is a well known phenomenon with sigma-delta type DACs but its very interesting to see it on an ESS device because ESS are the leaders in understanding this kind of artifact IMO."
Mathematically? Yes. (But I hasten to add, I take the word of mathematicians on this, I'm poor at math.)
You appear to have misread his explanation.
Item 1: A 16-bit ADC with dither can encode signals of amplitude lower than that represented by the LSB. The signal never fades away as the level gets lower, it just gets buried in the noise. Exactly like analogue. Do you accept that, or shall I post a short tutorial?
Item 2: 16-bit dithered digital has a noise level of about -93 dB (a bit lower if we use shaped dither). 16 bit implies -96dB, but we add +-0.5 bit of dither. The important thing is that this is broadband noise (0 to fs/2, or 0 to 22 KHz for 44.1 KHz digital.) Now we record a 3 KHz tone at, say, -130 dB. It might appear that this is well below the noise floor and should be inaudible. But remember that the noise is spread across the whole 0 to 22KHz range. At any given frequency, the noise will be much lower. For example, if we filter out everything at the DAC output below, say, 2.9 KHz and above 3.1 KHz, our bandwidth decreases by a factor of 100. (20,000 / (3100 - 2900) = 100.) The amount of noise in that band is also 1/100 of the -93dB figure. If I've done my math right, this reduces the noise floor in that narrow band to the vicinity of -130 dB. So our -130 dB tone should be quite visible and audible in the filtered output.
Now our auditory system comes into play. Our ears have the equivalent of filters. They break up the range into bands and analyse the energy in each band, just like the filter example above. We also have discrimination: We can pick out signals that "don't belong". We can pick out the 3 KHz tone because it doesn't sound like the noise. We're hard wired to do this. You may have noticed if you listen attentively to white noise, after a while you may start to hear fragments of other signals (voices etc) in it.
Item 3: When he said "amplified", he meant that the output of the DAC has to be amplified to a level where the noise (and tone) becomes audible, not that the tone had to be amplified before encoding.
I was replying the same as Don Hills, but his explanation was better. So I removed mine.
Perhaps these FFT screenshots will help a bit along with Don's explanation.
The first two are -93 db white noise. One is a 128 bin FFT, and the other is a 65,000 bin FFT. You see by splitting the analysis into smaller bands each band holds less energy. Hence the level is lower. Think of the FFT as splitting the 20,000 hz in the plot into equal sized bands, then analyzing the energy in each band. With fewer bands covering more frequency the level of each band is higher. With larger numbers of bands splitting into smaller frequency bands the level is lower.
Next is where I digitally created a 3 khz tone at -135 db. I saved that file as a 16 bit file with shaped dither. Then reopened it for this FFT. You see the noise floor with most of the noise at upper frequencies. You also should see a thin line at 3khz. I could have used a large FFT bin size to bring it out more, but the display of the picture makes it hard to see. I also zoomed in on the next picture to show it better.
So though this is a 16 bit file with dither you can encode and recover signals below the -96 db noise floor. Now even amplified I could not hear this. However at -115 db for the tone I can. At -120 for the tone I can still hear it faintly. In both cases I need to amplify it 50 db which means I hear a low pitched whooshing noise and a tone buried in it.
Of course the point isn't we hear these effects without amplifying, but merely to show with dither digital has the ability to record below the basic noise floor of the least significant bit. Since our hearing splits things into bands as Don Hills explained we can hear into a noise floor somewhere between 10-20 db if the noise floor is mostly flat. With some noise shapes we can hear further into the noise.
Perhaps these FFT screenshots will help a bit along with Don's explanation.
The first two are -93 db white noise. One is a 128 bin FFT, and the other is a 65,000 bin FFT. You see by splitting the analysis into smaller bands each band holds less energy. Hence the level is lower. Think of the FFT as splitting the 20,000 hz in the plot into equal sized bands, then analyzing the energy in each band. With fewer bands covering more frequency the level of each band is higher. With larger numbers of bands splitting into smaller frequency bands the level is lower.
Next is where I digitally created a 3 khz tone at -135 db. I saved that file as a 16 bit file with shaped dither. Then reopened it for this FFT. You see the noise floor with most of the noise at upper frequencies. You also should see a thin line at 3khz. I could have used a large FFT bin size to bring it out more, but the display of the picture makes it hard to see. I also zoomed in on the next picture to show it better.
So though this is a 16 bit file with dither you can encode and recover signals below the -96 db noise floor. Now even amplified I could not hear this. However at -115 db for the tone I can. At -120 for the tone I can still hear it faintly. In both cases I need to amplify it 50 db which means I hear a low pitched whooshing noise and a tone buried in it.
Of course the point isn't we hear these effects without amplifying, but merely to show with dither digital has the ability to record below the basic noise floor of the least significant bit. Since our hearing splits things into bands as Don Hills explained we can hear into a noise floor somewhere between 10-20 db if the noise floor is mostly flat. With some noise shapes we can hear further into the noise.
Last edited:
Don,
I would like the tutorial because I understand how it is possible when already in the digital domain and with small signals, but not specifically when the 16bit ADC is converting a complex soundwave (say symphony crescendo of an orchestra) that has true dynamics of over 100db.
The problem is the PCM stream must rely upon n-bits encoding, not an issue until you are talking about huge amplitude complex sounds with many harmonics; such requirements are pretty small as I have only seen a few hirez records reviewed-measured that have been measured with the requirements of needing over 100db (if taking ALL the frequency range into consideration up to say 15khz).
That said this is something none of us should be losing sleep over because IMO if differences are being picked up it has nothing to do with extreme dynamic range potential of LPs/CD/24-bit.
IMO of course
Cheers
Orb
I would like the tutorial because I understand how it is possible when already in the digital domain and with small signals, but not specifically when the 16bit ADC is converting a complex soundwave (say symphony crescendo of an orchestra) that has true dynamics of over 100db.
The problem is the PCM stream must rely upon n-bits encoding, not an issue until you are talking about huge amplitude complex sounds with many harmonics; such requirements are pretty small as I have only seen a few hirez records reviewed-measured that have been measured with the requirements of needing over 100db (if taking ALL the frequency range into consideration up to say 15khz).
That said this is something none of us should be losing sleep over because IMO if differences are being picked up it has nothing to do with extreme dynamic range potential of LPs/CD/24-bit.
IMO of course
Cheers
Orb
Similar threads
 | Steve Williams Site Founder | Site Owner | Administrator |  | Ron Resnick Site Owner | Administrator |  | Julian (The Fixer) Website Build | Marketing Managersing |