I would gladly pay for the shipping on any working pairs you would like to send me.
But if we could get back to the more interesting question, how about "better" speakers than NS-10s? You know, like mastering engineers actually use. Or something closer to that. It can be some speakers that you like. No problem whatsoever with that.
I'm far far more concerned with the room/speakers (headphones) when driving the limits of detection.
Insofar as I've tried, I cannot the difference between your test files Mark. Then again, my equipment is rubbish (although I'll probably be picking up an akm4495 + 49720 for headphone use soon), so have what you will. I couldn't hear the -60 dB sousaphone mix-in in Bill Waslo's audibility tests as well. So let's err on the side of me being utterly tone deaf. Agree with Scott that the "good equipment" clause is a very easy route to "no true Scotsman" type arguments, which I'm wont to avoid.
The paper I read that have tested 16/44 vs 24/xx have only highlighted that audibility at that margin (with what seemed to be pretty good controls/equipment/personnel) are below the level of their testing schema: http://www.diyaudio.com/forums/digi...rez-you-cant-hear-difference.html#post4487899
Subgroup analysis is the best way to take null tests and make them positive, and at least this paper points towards that being the case. And at p < 0.05 on three sub-tests of the suite (remember out of all 48 permutations, only 3 passed this threshold), there's more than enough room for pure chance to drive that identification. Or a faulty down-sampling algorithm, as only one was tested (and unfortunately not described). My memory might be faulty here, however, and this test might be 24/44 vs 24/88 -- I'll have to reread the paper.
There's also this: AES E-Library The Audibility of Typical Digital Audio Filters in a High-Fidelity Playback System, which I'll give a read at home, when I have access to my old university's journal access. 🙂
Moral of the story is that even in trained professionals with pretty careful controls, these differences aren't materially showing up. Not that I should recuse myself from earnest testing to see if *I* hear the differences, but points towards an extremely small effect size, very possibly below our collective hearing potential. Any claims that it's easy to tell 16/44 LSB dithering or purported clear audible differences between 16/44 and 24/xx files otherwise identically prepared is going to meet with me looking for technical fouls first, and the ability of the listener about 50th.
Insofar as I've tried, I cannot the difference between your test files Mark. Then again, my equipment is rubbish (although I'll probably be picking up an akm4495 + 49720 for headphone use soon), so have what you will. I couldn't hear the -60 dB sousaphone mix-in in Bill Waslo's audibility tests as well. So let's err on the side of me being utterly tone deaf. Agree with Scott that the "good equipment" clause is a very easy route to "no true Scotsman" type arguments, which I'm wont to avoid.
The paper I read that have tested 16/44 vs 24/xx have only highlighted that audibility at that margin (with what seemed to be pretty good controls/equipment/personnel) are below the level of their testing schema: http://www.diyaudio.com/forums/digi...rez-you-cant-hear-difference.html#post4487899
Subgroup analysis is the best way to take null tests and make them positive, and at least this paper points towards that being the case. And at p < 0.05 on three sub-tests of the suite (remember out of all 48 permutations, only 3 passed this threshold), there's more than enough room for pure chance to drive that identification. Or a faulty down-sampling algorithm, as only one was tested (and unfortunately not described). My memory might be faulty here, however, and this test might be 24/44 vs 24/88 -- I'll have to reread the paper.
There's also this: AES E-Library The Audibility of Typical Digital Audio Filters in a High-Fidelity Playback System, which I'll give a read at home, when I have access to my old university's journal access. 🙂
Moral of the story is that even in trained professionals with pretty careful controls, these differences aren't materially showing up. Not that I should recuse myself from earnest testing to see if *I* hear the differences, but points towards an extremely small effect size, very possibly below our collective hearing potential. Any claims that it's easy to tell 16/44 LSB dithering or purported clear audible differences between 16/44 and 24/xx files otherwise identically prepared is going to meet with me looking for technical fouls first, and the ability of the listener about 50th.
Last edited:
Without knowing more about what equipment they used (Windows sound engine?), etc., listener distance from speakers, room reflections/acoustics, etc., it would be very hard say if a particular study is likely credible or not.
However, I would say if the experiments relied exclusively on equipment specs and measurements, and didn't have anyone on the team who could reliably hear the difference and who could verify performance of the test setup, then I would probably be skeptical of the outcome.
I say that from practical experience. I have some pretty good headphones that I thought should be revealing enough (Sennheiser HD 600), but it turned out they weren't. I double checked with other transducers and the headphones appear to be lacking. Experiences of that nature make more more cautious than I might otherwise be regarding some of the research that has taken place so far.
Of course, its not just hearing research. I read an earlier post by jcx listing missing information from a paper JC referenced. Many researchers don't want to publish test details and/or raw data. It makes it very hard to go back later and try to find out exactly what was done.
For original recording and mixing its useful to have something that will highlight problems. You want to hear the instrument dropped in the background or the missed cue by the 3rd french horn. Your not making subtle balance judgements at that phase, your culling problems out of the stems or takes before you put hours into getting the right blend. However you do want to hear the subway rolling by but how many customers will hear it anyway?
A more balanced 'sweeter" sound would not be that helpful.
Actually, common use of stems in mastering has been fairly recent. However, more and more mastering engineers are accepting at least limited stem work.
Back when NS-10s where more widely used for all of mixing, there was much less use of stems, and in fact NS-10s were very good for making very subtle balance adjustments in the frequency range where hearing is most acute. They still are, but there are other better sounding speakers available now that are also good for that and that don't have the shortcomings associated with NS-10s.
Other than that, I liked that you raised the issue of how mixing work can be divided into two parts. First would be doing things like fixing problems in tracks, and comping together tracks of the best takes. The second stage is where artistic decisions are made about how to present the tracks that have been recorded and cleaned up as much as possible, into a coherent, musical sounding composite, the "mix".
Most mix work is still done prior to production of stems. Stems typically consist of only a few already mixed tracks, perhaps one with lead vocals, one with background vocals, maybe one with rhythm guitars, one with lead, etc. The stems are made by muting all the tracks in the completed mix except the ones that go a particular stem. The selected tracks for a stem are mixed down at the same volume and balance as there were in the completed mix. So in theory, if one plays back all the stems at the same time with no playback level changes, the result will be the exact already completed mix.
Then, if a mastering engineer decided the lead vocal needs to go up 0.5 dB, it can be done by turning up the stem level just for the vocal. Prior to stems, the mastering engineer could try to use things like EQ and/or mid/side processing, to fix the problem or if nothing worked, send it back for a remix.
It has also become apparent that mp3s and similar formats often require different mastering than for CD format. That is primarily because perceptual encoders are somewhat unpredictable. To get an mp3 to sound as close as possible to a CD often requires some trial and error to get the best results. Stems make it easier to make limited adjustments to completed mixes for purposes like that.
However, having said all the above, stems are continuing to be more commonly used, and mastering engineers are increasingly being asked to do more mixing, which many of them don't like. For one thing, part of the purpose of mastering is to have someone do it who didn't do the mix. A fresh set of ears is good idea. It's like a writer having a proofreader, something that's hard to do for one's own work, at least until sufficient time has elapsed.
Last edited:
One really great way to know how good a system is is to play back tunes to the guys who were there when sad tunes were being mixed or mastered. I did that just a month ago.
The audio in LIGO I refer to is Rainer Weiss was into trying to reduce record noise on old shellac discs. Knowing he needed more education he ended up at MIT and back as a professor in a round about way(women). He did the noise calculations to see if it was even possible to measure and then secure the $ to do such a thing. Makes audio look pretty silly compared to these machines and their electronics.
What is "good" reproduction can also be related to the necessities inherent in the reproduction itself based upon ones interests / tastes. Let me elaborate as it may illuminate Scott's comments.
For those of us (few, admittedly) who have great interest in ethnographic recordings or other "fringe" recordings which may have seen limited release or were recorded under less than ideal conditions things can be a bit different for example. These recordings in some (but certainly not all) cases do not have the following to justify quality remastering or reissue, or even after restoration work it's simply not ideal / what one would consider hifi. Some of these used less than optimally configured portable equipment used on site at a remote location, and so on.
A substantial portion of my LPs contain these sorts of recordings.
I think it's unfair to sequester high fidelity away from making all types of music and recordings accessible to people- a good system can make even these very poor recordings come alive in a way a standard system cannot and place one on a closer "plane" with the artists.
In many ways I relate the disregard for lower quality recordings by music aficionados in the same category as being adverse to watching a foreign film for having to read subtitles- it's a bit on the closed minded and masturbatory side of things.
Anyhow, ones needs for those recordings are very different from those interested in classical works or blue note LPs or rock records from The Beatles or The Dead which had larger caches of money, equipment and engineers at their disposal at the time to capture higher quality recordings. Similar is true for the higher baseline of quality equipment for those recording and listening to all styles of contemporary music.
Greater challenges are presented to someone who appreciates all sides.
Speaking for myself, I'd opt for some level of "colorful distortion" or "guesswork equalization" or switching to a tube preamp for a smoother delivery over suffering through "straight wire with gain" playback of something recorded on a beat up local mass market tape machine in the desert- especially when the music is of interest to me.
Is suffering good? Perhaps, in some instances. But what I know certainly is that I'd like to avoid suffering in the moments I'm taking for myself for musical enjoyment.
In this instance if the recording quality is low even a straight wire with gain is not reproducing anything remotely pristine anyhow.
My personal view is that a good system (for me) is one thats not only a technical achievement for optimal recordings, but also something versatile that can help us glean real enjoyment when provided with less than ideal input.
One can do their best to approach a technical pinnacle; but also not be so stubborn as to suffer within the confines of their own perfection.
Even John uses TDK pots and remote controls once in a while; after all he's gotta make a paycheck and has to sell these things to actual real life humans beings. Is that not a good reproduction system because it doesn't have a shallco stepped attenuator and isn't hyper-minimalist? Well, I think one would be pretty misguided to suggest that.
I think most people, including perhaps some of the most technically astute engineers within this thread, make what I would call "minor preferential adjustments"- perhaps their cart or cart loading, speaker placement, or some convenience or aesthetic compromise such as "I don't want bass traps on my wall or "my listening position isn't perfectly aligned because there's a walkway in between."
As is often the case, life does not present us with ideals. We have to make sacrifices and find our ideals in how we manage the imperfections. Sometimes that's 10W sets and horns, sometimes it's 200W JBLs. What is right is open to infinite debate; as the many pages of this thread show.
Claiming to know of an objective good for everyone in all rooms and situations sounds a bit like one of the bullet points on the WHOs genocide precursors checklist.
Another thing that is often the case, is many people will vehemently disagree with me 🙂
This "suffering through "straight wire with gain" playback" is a myth, IME. If you suffer when listening to something claimed as straight wire with gain, then something is definitely technically wrong with the thing. If there is too much noise in the original recording, better use tone controls than distorting equipment.
Similar for "colorful distortion", there is no colorful distortion, only distortion.
My opinion and experience as well.
A tone control would be considered a distortion of the original signal, as well as the introduction of color, no? Tone controls are not perfect and imagining what "accurate" would be on a historical recording and attempting to reproduce this with tone controls adds two layers of error: technical and human.
What I meant is that absolute precision in reproduction of something with imperfections yields the playback of imperfections. So, one can adjust for imperfections. Is that good or not good is endlessly debatable. What is considered an imperfection is again further debatable.
Coming to an objective agreement on what is "real" or "accurate" in this regard is almost an exercise in futility.
I was commenting on someone's use of semantics that are philosophical and/or moral and how they are not useful, not making a case for my own preferences.
What I meant is that absolute precision in reproduction of something with imperfections yields the playback of imperfections. So, one can adjust for imperfections. Is that good or not good is endlessly debatable. What is considered an imperfection is again further debatable.
Coming to an objective agreement on what is "real" or "accurate" in this regard is almost an exercise in futility.
I was commenting on someone's use of semantics that are philosophical and/or moral and how they are not useful, not making a case for my own preferences.
It was also a suggestion that depending on ones tastes, recording quality or time period can be a more dramatic influence in terms of introduced noise or distortion of the original "live experience" than the associated playback equipment, in which case what is "good" or "pleasing" for reproduction, on the whole, is necessarily subject to personal interpretation as we do not have the original musicians on hand in a consulting capacity..: well, at least I don't.
Again, not a global statement but an illustration of a particular scenario.
Again, not a global statement but an illustration of a particular scenario.
I think that in most of these discussions here 'distortion' is meant to be non-linear distortion, i.e. adding harmonics due to a non-linear transfer function.
The term 'linear distortion' has been used to denote non-flat frequency response which includes of course tone controls, EQ, loudness etc.
I assume you mean distortion in the latter sense, which for many people is not distortion per se.
Jan
The term 'linear distortion' has been used to denote non-flat frequency response which includes of course tone controls, EQ, loudness etc.
I assume you mean distortion in the latter sense, which for many people is not distortion per se.
Jan
I think you are mixing up 'good way to get marketing quotes' and 'any form of reliable testing' unless said guys also pulled out their original master tapes and A/B'd them. In which case gigantic Kudos.
Can I just confirm something. Your 24bit vs 16_dithered tests, can you hear the difference in your domestic system or just on the 'mixing setup' with NS-10s (and sorry for any misassumptions if the NS-10s are your mains for muscial enjoyment just no one I know listens to them for enjoyment). You also infer that there are headphones that are resolving enough. Can you recall which those are (not that I can afford anything better than my Koss).
Aside: there is an interesting shootout Bob Katz has been doing on various headphones over on innerfidelity. What is (for me) really fascinating is that at the end he and Tyll disagreed on one of them, so they compared measurements and found that the QC on some of these high end cans is dreadful.
Agreed that distortion is unlikely to be the most desirable solution for improving ethnographic recordings. Even if one did want added distortion, adding the exact same distortion to every recording would probably not be the best way to go. An accurate reproduction system, with some real time VST plugins would probably allow much more versatility, with the option of accurate playback always there.
Also, agreed that true straight wire with gain ought to be a good thing, at least for most purposes.
Only quibble is with the generality of "colorful distortion" doesn't exist thing. In very low doses, some people do seem to like small amounts of 2nd and 3rd harmonics, although it can be easily overdone. And, for applications other than reproduction, it is essential for certain types of electric guitar sounds.
Sorry for the long quote, but I think there is an important point here. Historically 'audiophiles' and 'music lovers' were considered to have very little overlap. My feeling is that theses days the centre of the Venn diagram is a lot fatter than some people might want you to think. For me the music comes first and based on the comments on here it does for the majority.
I've not seen many 'music afficionados' on here. I can only really think of one. And there is a general acceptance of defeatable tone controls.
I suppose I'm saying you are amongst friends here 🙂
Bill, currently I do have NS-10s set up in the living room. I have other speakers that in many ways sound better, but not as revealing. It would be nice to be able to hear details and have good sounding speakers at the same time. What I would like to be able to do would be to listen and compare some prospective new speakers in a quiet showroom using some hi-res files, but there is no such place to do that in my locale, at least that I am aware of.
Regarding headphones, until fairly recently I thought the HD 600 were pretty good. And actually, there is a possibility they are better than my current impression. When they failed to resolve what I wanted to listen to, I attributed the failure primarily to the headphones, but they were fed by the DAC-1 headphone output. It turns out that DAC-1 line out specs are better than headphone out specs, so it would probably be worthwhile to double check with another headphone amp.
However, I have used Etymotic earbuds at times and found them pretty revealing, although I didn't like them. They sounded worse than NS-10s to me, but for different reasons. They have a wiggle in the frequency response intended to compensate for ear canal resonance. However, they don't seem to match my particular resonance, so FR sounds very odd. Also, they have to be very tightly sealed to hear bass well. Etymotic says they are flat everywhere except the ear canal compensation thing, which I don't dispute. But the tight seal needed for accurate bass reproduction can be uncomfortable.
Yes correct. My vocabulary and knowledge in this area isn't very vast to say the least. I used the term to delineate "not original", or adjusted / altered.
I was attempting to illustrate the futility in defining what is "good" objectively, as it's a word that implies subjectivity, as well as a moral standing.
As a preamplifier designer, especially within this thread, of course your goal is the ideal- ultimate transparency.
I think good is a term better reserved for flattering domestic pets and so forth.
Maybe too in a multinational forum words get minced up a bit more than usual.
But also perhaps I'm in over my head.
Last edited:
Thanks. I use etymotics at work and agree on the seal. The rubber seals just don't work for me but the foam tips do. However they seal too well, so drinking a cup of coffee whilst wearing them is an odd experience. Also wearing them for the odd commute to london is very odd as you walk along and hear every clunk. I find Comply tips to be a good compromise.
I have experimented with adjusting for the Harman curve as per some papers Tyll published (will try and find the link tonight) which does help a bit*. He did a load of additional measurements at the Harman labs a few months back but sadly they haven't been fully analysed yet.
*Note tho that critical listening is not possible whilst working but nice to work with office noise blocked out.
I have experimented with adjusting for the Harman curve as per some papers Tyll published (will try and find the link tonight) which does help a bit*. He did a load of additional measurements at the Harman labs a few months back but sadly they haven't been fully analysed yet.
*Note tho that critical listening is not possible whilst working but nice to work with office noise blocked out.
- Status
- Not open for further replies.