Defining High-Resolution Audio
The posts over the weekend described what high-resolution audio isn’t and what it is. Today, I guess I’m going to offer my definition of high-resolution. For those that have been reading these posts since the beginning (which started on April 2, 2013), might remember a post from April 4th. I launched this site with a definition. I think it’s still pretty accurate…which is pretty good considering all of the months that have gone by…but it might need a little tweaking to make it to the masses. The only thing that I’ve changed is the name. I was (and still am…) in favor of HD-Audio but I have moved over to High-Resolution Audio because it seems others believe that’s a better name. Don’t ever say that I’m not flexible.
Here’s what I wrote last year:
HD-Audio is a recording that has been captured during an original session using equipment capable of matching or exceeding the capabilities of human hearing. If the generally accepted measure of the human auditory system includes a frequency span of roughly 20 Hz to 20 kHz and a dynamic range that tops out at around 135 dB, then a recording system would need to be able reach these specifications to be considered HD. In the world of PCM digital recording, this would translate to at least 48 kHz and 24-bits. Given there is some evidence that higher frequencies may impact our listening experience and that moving to 96 kHz has advantages for equipment designers, it’s seems reasonable to adopt 96 kHz as the minimal HD sampling rate (I would accept 88.2). As an engineer/producer creating new HD tracks, I choose 96 kHz/24-bits as the minimal specifications to achieve HD-Audio.
This applies to the source recording…when the musicians are (or were) actually in the studio laying down their parts. It does NOT mean that all files called “high-resolution” tracks are actually qualify Sadly, most of what is being sold through the different HD download sites are no better than the CDs that you already have and they may be worse if the wrong masters are used. The sound may be different but the actual fidelity is no better than before.
Here’s a mass-market version of the definition:
High-resolution audio (or HRA) means a source recording was made that meets or exceeds the full capabilities of human hearing, which means a frequency range of 20 Hz to 20,000 Hz and a dynamic range of 135 dB.
Let’s parse this statement and see if it really makes sense in the context of the last couple of days of posts.
1. We’re talking about original recordings that can take advantage of the fidelity (regardless of format) of our ears. I don’t honestly care if the tracks were made on analog tape, DSD, vinyl LPs, wire recorders or PCM…but they must be capable or have the potential of delivering 20-20,000 Hz AND a dynamic range of 130 dB.
2. How does this definition do with regards to sources and delivery formats and specs? If the source is made with fidelity in mind, say at 96 kHz/24-bits PCM then audiophiles can get the benefits AND lesser fidelity files can be derived from it for iTunes or CDs or anything else. If the delivery container has “Hi-Res” specs but the source doesn’t, then the delivery is not HRA. This is where proper provenance information will be needed…which may be a real challenge.
Give me your feedback. I feel pretty good about this as a simple metric for HRA
Mark,
I agree with you about your dynamic range specification for HRA. Anything less than your 130 dB point will let CD’s, LP’s, Analog tapes, MP3, cassettes etc. drop out. None of those sources can provide that high of dynamic range, even if they contained HRA recordings. I like that.
I don’t agree with you on the frequency response number. An open reel tape recorder or a CD and some LP’s can contain frequencies up to 20 KHz. The HRA frequency specification should be at least 24 KHz. That would reserve the recording source to be at least a 48 KHz sampled digital signal or a fine-tuned, highly polished analog recording on some kind of souped-up analog tape recorder (perhaps 1″ tape running at 30 IPS).
Another problem with the simple 20-20KHz specification is that it is not referenced to a +/- dB limit range. Then we may go back to the old days where an amplifier was rated at 20-20KHz, but +/- 12 db not mentioned.
The masses probably won’t relate to a +/- dB limitation, but without it, you leave yourself wide open.
Your thoughts on this would be appreciated.
I would suggest that the minimum frequency specification be 40 kHz, but there are plenty of knowledgeable folks that just brick wall hearing at 20 kHz. I don’t. I know that a room of musicians will produce frequencies the exceed 20 kHz…so why not capture them and put them back in the room. Remember this is a minimum specification.
We need to have 20-20 kHz as the human metric and show that the equipment chain meets that standard…including accurate measurement of speakers. I’m trying to avoid too much techno jargon.
I have some trouble believing that a vinyl LP can deliver up to 24 kHz…I hear different from my disc mastering friends. I’m sorry I’m going to miss Doug Sax tonight speaking at the AES meeting. I would ask him.
This definition looks good. I wonder if one could reduce the dynamic range requirement from 135db to the range that one would encounter in a real listening environment that was optimized to be “quiet”. It is very difficult to get the noise floor of a quiet listening environment below 10-20db??? If this is the case, that would allow the dynamic range to be reduced to about 120db. If accurate, would that help with potential agreement/adoption of this definition?
I wholeheartedly agree with your definition. I also believe that HD or High Res music sites should have to list not only the fidelity of the “master” but where that “master” was sourced and have verifiable proof of such.
The source AND the delivery specifications need to be identified. That’s what the HRA Database at HRA Planet.com will have soon.
Mark,
Curious about the 130dB standard.
With -16dBf as a zero VU, doesn’t this put most of the 130dB below the -16dBf ref level, and if so at what level in a normal (perhaps that should read common) listening) environment are levels recognized by the ear over typical ambient sound?
I understand we have that lovely control “The Gain control Potentiometer”, and this can raise the floor above ambient, but how much dBm at the top will one need, to hear that fingernail scrape on the cello at -114dBf?
Mind you, I crave to hear as much of this range as possible.
Thank you,
John Chase
John, leave aside how our equipment and metering handle or reference the dynamic range of a selection of music. The range is defined in acoustics terms…what human ears can “hear” from the quietest butterfly sneeze in an anechoic chamber to the sound of an F-15 with afterburner blazing ont he deck of an aircraft carrier. Then there’s the music range from pianississimo to triple forte.
Setting gain stages and playback levels in the studio or in a home theater or from a Pono player is another thing all together.
I believe the average listening room has a noise floor of around 40-50 dB SPL…so I recognize that the realities of individual spaces will matter a lot.
Precisely my curiosity!
Since listening environment and listening needs come into play so much, just how much dynamic range will be utilized by the typical user?
Not making waves here, asking questions and hoping the bottom line of resolution is at least 96/24, which is what my wish has been since it has been a possibility.
When I was young, we had a band that played a lot of parties, and when any one played people seemed to enjoy listening to the music.
Except for concert events and shows, there seems to be a lot of talking at live music events.
Music appears to be more of an ambient noise product for the average listener, and the ability to set a level that varies by a couple of dB is convenient and unfortunately, continuously loud.
So, having read your posts over the last number of months, I believe you are looking for a ‘Recorded in High Resolution’ tag/statement for recordings, giving their provenance.
From the consumer’s viewpoint if it is marked as ‘Recorded in High Resolution’ and the download delivery format is 24/96 or above on Blu-ray disk you should be assured that you are not getting Analog Tape/CD quality recordings repackaged as High Resolution Audio.
Plus moving the provenance statement back to the Recording, it still allows the statement to used by the Labels on CDs/Vinyl (non-High Resolution Audio formats) differentiated as from a High Resolution recording process, and therefore still better than before, for the consumers that haven’t upgraded their reproduction equipment to facilitate High Resolution Audio.
Thanks,
Simon.
Recognition that both the source recording format AND delivery format specs are valuable is going to cost some marketing points for Pono and the others.
Nice Dr. Mark. By keeping the specs for HRA *container and format free*, you’ve made allowances for future technology improvements. I think the whole HRA movement, guided in part by earnest efforts like yours, will finally removed most techno mumbo-jumbo behind, and let the quality and emotional connection of the music come through.
Thank you!
Thanks Barry…we’ll see.
For me the problem is the what is the definition of “resolution”? I am a scientist, so I am happy to talk about the resolving power of a telescope or a microscope or a chemical analysis. But when you talk about resolution in HRA exactly what is being resolved and does it only apply to digital systems?
For analog systems I would understand that resolving power might mean we could plainly hear all of the individual instruments in say a Piano Quintet or even a small Chamber Orchestra. Obviously this would apply to both Analog and Digital music reproduction – music is analog after all. Is this what the R is in HRA?
I have a couple of recordings of ‘Kind of Blue’ – who hasn’t I hear you say and I know that on one of the tracks I can hear the brushes being dragged across a drum on a 180g DMM LP and that I just can’t on the CD version. I suppose I should now buy a 24bit version – but I won’t. So what does this prove?
Anyway sorry to ramble on, I love the field of hifi but I also hate it at times. When you hear of people selling bits of magical wood, or blue marker pens or silver ethernet hookup cables, all at vastly inflated prices, then it makes my blood boil. But I just love listening to music from Wagner to the Dead, from Pink Floyd to Sandy Deny, from Vivaldi to Glass, from Dylan to Dylan – I just love it and I just pray for some sanity in the HiFi industry, but I ain’t holding my breath.
Bill
Resolution in our digitized image and sound world is perhaps very different than optics or chemistry. It’s a specification that derives from the sampling/quantizing requirements associated with images or waveforms. The more samples and amplitude points the higher the resolution. As for hearing individual instruments or low level elements in a texture of loud sounds, that’s more about the miking and recording techniques.
HiFi is a unique category. The best we can do is just listen to what turns us on.
So, it cannot be applied to a purely analog format then?
It absolutely can…once we’re past the digitizing stage.
Now I am totally confused. You say ” derives from the sampling/quantizing requirements associated with images or waveforms” but in analog reproduction surely there is no .sampling/quantizing .malarky, or am I missing something?
Apologies I didn’t read your post properly. But, if the recording NEVER goes into the digital domain then the word ‘resolution’ as used in HRA does not apply. So it seems to me that to say analog tape is not HRA is kinda meaningless.
Believe me I am not a dyed-in-the-wool analog hippy but I have a huge vinyl collection, but right now I am only listening to digital because of two things:
(i) I recently bought a NAIM Unity box and a pair of KEF R700 speakers and this combination sounds brilliant. Really bad CDs (unfortunately some early Airplane stuff for example) still sound bad but so much sounds brilliant. I was listening to a CD version of Voodoo Chile, the live version from ‘Electric Ladyland’ last night and it blew my socks right off and so do many other CDs. RESULT!
High bit stuff sounds even better.
(ii) The second reason is that the Air-bearing tonearm on my beloved Galibier turntable is playing up at the moment. Must get this fixed.
The reason that analog original recordings when digitized need to be classed is that the entire premise for Pono and other downloads are based on the ability to move these older tracks to High Res status…which is not true.
Bill…it not possible to talk about resolution with regards to audio or images prior to the digitization process. Whether it’s 72 pixels per inch or 192 kHz samples per second…we measure resolution after the slicing has been done.
Up to a point, Lord Copper.
I don’t agree with your “completist” approach of insisting that, to be called HRA, the recording and replay chain must be capable of matching or exceeding the capabilities of human hearing. In other words, of recreating the full sound generated by the original performance, including the inaudible parts.
Leaving aside the risk that harmonic interference between audible and ultrasonic sounds will create audible artefacts, a matter for amplification and filter designers to address, the fact remains that if sounds are inaudible then you can’t hear them, full stop. Their incusion or exclusion in the soundscape as presented to the listener is of interest only to a bat.
I know you’re at odds with the whole notion of ultrasonics. If those frequency components are in the room where the musicians performed…then the ultimate in “fidelity” would record them and reproduce them. It’s up to the equipment chain to handle the ramifications of interference etc. Why categorically rule out the possible of ultrasonics when it costs nothing more to include them?
Dr. Mark WALDREP.I totally agree with you on your definition of distints audible sound qualities.I am not an audio expert but it’s not hard to understand what the technological truth as you have explained.Congratlations and GOD bless you.
Thanks!
Please convince the task force to stick with this definition and we are all set for a new era for audiophiles.
This will be a very challenging task. I had one of the board members in the studio yesterday with a couple of gentlemen from a large Japanese consumer electronics company. I played a few of my best tracks in surround and stereo…they got it.
From ARGENTINA this is my feedback for your metric HRA. Excelent work Mr. WALDREP.Thanks for all the info that you give to the lovers of real audible sounds.From Mr. EDISON and PHILIPS/SONY in 1982 CD makers,that is the past, is the same if you have an a sand clock in a digital era.The best is now starting from 96 KHz/24 (also as you said 88.2)Thanks again.GOD bless you.
It’s going to be a very long haul to get consensus on the 96 kHz/24-bit PCM minimum standard that I’m pushing for…we’ll just have to vote with out dollars.
In reference to Roderick’s comment…
The funny thing about high order harmonics, is when these overtone structures combine, they can form harmonics that are well within the hearing range of the ear.
Loosing these due to either poor recording technique, conversion processes, etc. can leave you with a reproduction that sounds flat compared to being there.
As an engineer, I love it when a reproduction can transport you to the event.
When it happens it is a great experience.
I like the definition. Whether we use 48K or 96K doesn’t really matter too much to me, but I agree that if there is something there, then reproduce it. Storage space is cheap and getting cheaper. Possibly download bandwidth may be an issue, but I suspect not in the future.
I would put an octave safety margin around the 20-20k human hearing frequency range — so 10 Hz to 40 kHz.
Dynamics: take the background noise level of a top recording studio (25 dB?) and subtract 10 dB, then at the loud end take the 105dB peak level that is expected of a reference grade home theatre and add 10 dB. So that makes 15 dB to 115 dB, or 100 dB range.
And the 100 dB performance available over the entire 10 Hz to 40 kHz range.
This way, HDA has a comfortable safety margin above any reasonable expectation of what is audible, instead of being exactly perfectly just sufficient.
*Sarcasm alert: I’m looking forward to loudspeakers that can deliver 115 dB from 10 Hz to 40 kHz!
I’m not far from you on these…but I believe it’s more important to avoid the realities of a specific environment and stick with the range of human hearing rather than expectations. There are DACs and amplifiers that can deliver 130 plus worth of dynamic range. And there are recordings that get close…including my own. It’s true that the speaker folks are the limiting factor right now. But I think it’s more important to get things right at the other end of the production chain.
Hi Mark,
In spirit, I agree with your definition of HRA being that which at least captures the full capabilities of human hearing.
But in practice, 130dB is unobtainable, so how can any HRA initiative based upon it succeed?
Everything you’ve ever recorded would be disqualified, and that would not be appropriate with respect to your intentions. Right?
And I also suggest that a dynamic range less than 130dB would be quite acceptable for HRA’s intended purposes of delivering music and theatrical sound tracks. Commercial entertainment will not be played back at levels anywhere near 130dB SPL peak, and such releases will not leave headroom for the F-16 jet.
Perhaps it would be better to define the minimum resolution specs. that would divide Standard Definition from High Resolution. This means agreeing on defining the limits of Standard Definition, and agreeing on a margin above which would define HRA.
I would propose the limit of Standard Definition to be 48kHz sample rate, properly dithered 16-bit resolution. This is approx 24kHz bandwidth, and at 114dB A-wtd dynamic range with aggressive noise-shaped dither. On second thought, 114dB I assume is pushing into the realm of common 24-bit recordings and is rarely achieved in 16-bit, so how about we relax this limit to, say, 110dB?
But unfortunately, it may be nearly impossible for the interested parties in analog recording to agree what are the limits of analog recording and whether or not its capabilities may cross the threshold into HRA. But even for the analog camp, I’d say it’s worth a try to propose the SD limits as 24kHz BW, 110dB A-wtd dynamic range. If their tape is measurably better than this limit (state-of-the-art 2″, 30ips, Dolby SR?) then by all means let’s consider this level of analog to qualify for HRA.
Regards,
Joel
Mark,
While I agree with requiring that original session recordings be high resolution in order for the released recording to be considered high resolution, there are some special cases where this should not be required. Some music performances include elements that are not high resolution. For example, Respighi’s Pines of the Janiculum, written in 1924, ends with a recording of a nightingale played on a Brunswick Panatrope record player. I would allow a mix that is composed of high resolution sources except for the nightingale recording specified by Respighi to be considered high resolution.
I’ve actually recorded the “Pines” in high-resolution and mixed it in surround as well as stereo. The birds in my recording were played from a tape supplied by the publisher. The sound is in the room and we recorded it. The entire project is then a high-resolution release.
Yes, I have that recording (AIX 80006), and I certainly don’t want your definition to exclude it from high-resolution. But, what if you had used the nightingale tape supplied by the publisher in the mix instead of recording the playing of the recording? For the Pines of Rome I would not do this, but for some types of music this does make sense, and I would want the release to still be considered high-resolution if all the “new” performances were recorded in high resolution.
I agree with you…but it ultimately could be that more than 50% of a project is made up of older samples of various things and then what do you do?
Record new high resolution recreations of the lower quality parts, and produce superb reproductions of the originals with great attention to the intent and artistic aesthetic of the original?
I have been listening to some of the HD Trax dowloads as time allows, and I like what I hear so far.
What I see on a set of VU/PPM meters, is an average of somewhere between 10-20dB between most tracks.
This is fantastic, and much nicer than the usual 2-4dB often seen.
Hi Mark,
I agree fully with the difinition as formulated, and especially as it seeks to set a valid quality standard. Although I share some of the questions regarding the usability of 130dB dynamic range in the common environments we listen to music in, etc., there has to be a univocate quality standard if we are going to pay a higher price for a product category that until now, has been mostly hype surrounded by loads of misleading information and ultimately false advertisement.
If I could add a humble suggestion, it would be that record labels at least included a list of the recording gear used for each recording as part of the credits, as well as the most significant steps mastering went through. I usually look for the information regarding gear used, and until now I have found that only a handful of labels detail the equipment they use, but also not consistently (like 2L, Caro Mitis, MA Recordings and Acousence).
When it comes to mastering, mixing and editing, it is highly unlikely that such a detailed and variable process from label to label could be easily or well documented, but it is not up to the consumer to perform a quality control function on products that are offered as guaranteed to be what they are said to be, like QOBUZ’s Studio Master Guarantee. Assuring that the conditions necessary to validate and legitimize that guarantee, and thus the correspondent and necessary steps of a quality control, have to be included in the costs of the product. Otherwise, the advertisement should warn the consumer that it is up to the customer to validate the Studio Master guarantee offered, which obviously – not to say ridiculously – defeats the purpose of a guarantee and renders it a mere scam.
Knowing what gear labels use and having access to the specs is one useful tool to at least be able to tell that their gear actually can record at HRA standards, but this would obviously not guarantee that the people operating the equipment use it correctly, nor that subsequent editing, mastering, etc processes won’t degrade the recording below HRA standards. A spectral analysis like the ones offered in this blog, with a simple guide on how to read them, would certainly be a good addition to avoid consumers reading huge labels and allowing them making a fair purchase decision.
The information needed to deem a recording as within the HRA parameters defined above, is not at all easy to obtain, to understand, let alone to demand from recording labels, but we can all see why the effort to offer a guarantee has to be in place, and why standards have to become regulatory. We can’t pay more for audio quality advertised as superior to CD quality, if the standards of these products can’t even guarantee the CD quality they say to have difinitively overcome.
Cheers
I’m in the midst of preparing a database of spectra and dynamic ranges with annotations…we’ll see if this can succeed.
That sounds great! That would certainly be a quality standard to go by, and although iTrax has a responsibility to the end consumer, it is of course primarily the labels who should provide this information and have this resposibility.
Maybe if few start providing real and valid specifications regarding audio quality of HRA downloads and provenance, others will feel the pressure to keep up and join better industry practices and transparency towards consumers.
Cheers
I read and accept that, even though apparently unproven, it is possible to sense and appreciate the influence of sound waves significantly above 20khz that may have been captured during the recording process.
But, even though I have a 24/192 and Dsd DAC, how can this sensing be possible if my amplifier and speakers are limited to 20-20khz?
It is doubtful that your amplifier has those limits but your speakers will definitely need to be very high quality to make a difference. And there may be in band impacts of higher frequency components (interference tones etc).
Thank you… Looks like more expense 🙂 So, I shall first consider buying some extended range headphones eg. HD800 and connecting these directly to my ( 24/192, 24/384, DSD 5.6) DAC to hear what my hi-res music should hear like before replacing my amp/speakers… I wonder whether hi-res amplifier/loudspeaker/headphone choices will become a topic in future audio event seminars? Are there any headphones with freq. > 50k? Time to research (again)…
Julian, it’s all about the productions themselves…not the formats.