@studio I’m following this thread and trying to learn from it so I appreciate that you are doing this.
But trying to make a close comparison of three tracks that run consecutively with no markers is difficult. When people post comparison tracks they usually post them separately so that we can run them in parallel and jump quickly back and forth to compare short sections. Just a suggestion.
Yes, I should have done that!
In my defense, I wasn’t even planning on using this 3 song YouTube video for this thread.
When I get back in my 
“la bore a tory”
I will make 3 separate versions for this experiment.
Just found this interesting to watch about the hi res DSD.
I guess people have been using it in their procedures to clarify older master recordings and reselling the recordings!
Great Video Explaining Digital Audio Basics
…and then there’s this:
well explained video. I did not know what DDD label on a CD was. I wondered that long time ago and never bothered to check again. Thanks!
A good take, though he makes a pretty big claim to be able to tell the difference btween 48k and 96k
Though I am not arguing against 96k, most pros today only work in 24bit 96k .
Agreed. I think blind-testing yourself is a really important exercise.
If an individual claims to be able to tell these subtle differences, but haven’t submitted themselves to a blind test, then I’m afraid I can’t really put much stock in their claims.
Back in the days of RecordingReview.com(not the current site that links from this address, btw) - where most long-term members here originally came from - we used to regularly test claims via blind testing. It was very illuminating, to say the least!
Interestingly enough, he finds himself a proponent of 48khz when he has aggressive drum sounds needing to be mixed!
His “opinion” is a subjective, Clinical vs. Punchy sound between the 2 rates.
He’s making a living with that technique so good for him!
Thanks for all your input btw!
I do believe in that video he does imply he did the experiment experience, then a quick edit switched to another point in the conversation.
Being able to trust someone at face value can be worrisome but it doesn’t always fall on the shoulders of mistrust either.
Some of my longest running clients in the corporate audio world were by sight unseen referrals.
Now, should I myself track my drums at 48khz because Colt Caparune said so? Of course not!
But it might give me an insight into a different methodology from what I normally do.
I’ll take the info and put it in my back pocket.Thanks.
What I take away from this topic is that 44.1 is pretty darn good already at CD quality, but 48 is better for video as well as good for audio. 96, although it would technically be far better, for me has two huge drawbacks-
Only my current PC has an i5 and 32GB RAM and a 250GB SSD and 500GB HDD, which already limits my setup to 32 tracks tops, but I have not tried that yet. All I know is my previous PCs and laptops could never handle well larger Reaper projects. I’d have to mixdown multiple midi to a single audio to exceed my hardware limitations, which was a bummer. So if I go 24/96 I suspect my happy PC will get a lot closer to its top performance. I would have to go 64GB RAM, which might not even be an option on my Lenovo. Plus I will gobble up hard drive space much faster and have to consider external drives for storage, and all that could impact performance as well. I’m also pretty sure all my VSTs are 48 tops if not 44.1 samples. So right now I feel I am good at 48/24, which Ingo and I just switched to after reading this topic. I also record mp3s at 320k whereas in the olden days everything was 128k because that was supposed to be as high as humans could hear, but I feel the higher the better for an mp3.
So for my quality of input I think my setup is very satisfactory, and 24/96 would be overkill, to put it mildly. But if I were Michelle or Christopher, I’d probably go 24/96 and downsize to 48 or 44.1 as needed.
In my younger hiking days, I didn’t like going up hill!
Hey folks,
Those are a few cherry picked quotes from this thread that, to me, sum up the hard line most audio engineers take when presented with an issue that impacts either their reputation or bottom line. Let me start by saying that I have been working as a pro audio engineer and manufacturer since before HRA or anything like that came into existence. My customers were and are mastering engineers, so they’re more fidelity–focused than your average audio pro. Unlike most pros, I also have one foot in the CE or consumer side of things, since CE (Consumer Electronics, as that industry calls itself) is what drives pro audio in terms of both hardware and distribution formats. So yes, I have very strong opinions on this, from decades of watching first 2x, 4 then, 8x sample rates and multispeed DSD be introduced. Also from working with some of the best “ears” in the biz. Sorry in advance for the rant but some posts in this thread either make me choke or chuckle, so here goes — limiting my reply to only two items:
• With recent generations of interfaces and storage, there is almost no excuse not to record at at least 88.2kHz. Yes, hi rez files take up more storage space than a 1x file but, when an 8TB hard disk costs $159 at your local Costco, is that really an issue anymore?
Since most folks can’t afford a fancy, standalone SRC product, it makes sense to record at 88.2 (2x CD rate) rather than 96k (2x video rates). Let the video folks SRC your 88.2 master to 96k since audio is subservient to picture anyway in any sound–for–picture application. And, 88.2 cleanly SRCs to 44.1 using even the most lame software. Besides, 88.2 takes up less space than 96k. 
• “Humans can’t hear a difference” is the stock rebuttal to going the HRA (High Resolution Audio) route.
Well, that’s sort of true but, as with all black & white blanket statements, the real answer is “It depends.” Our hearing mechanism is very, very “plastic.” We are constantly filtering, adjusting and adapting to our surroundings and the sounds presented to us. Studies by reputable researchers, like Jon Olive and others, have found that naïve listeners will not notice a difference with pretty much any fidelity improvements unless they are trained. The rise of MP3, Dolby Digital and other lossy codecs and fall of attendance at live, purely acoustic (no sound reinforcement) performances have trained the average listener to become used to pretty crappy fidelity. The latest example of that is Atmos, which is limited to 48k for delivery to consumers.
To everyone who says, “Show me the science!”, I say “Use that exquisite machinery build into your body!” Use and trust your ears, isn’t that what our craft is all about? Engineers and cognitive scientists are far from understanding and characterizing all the aspects and parameters that make up the human auditory mechanism. [A colleague, an über engineer who’s designed gear for Studer, SSL, Sennheiser and many others, asked me rhetorically last week, “Why is it that two pieces of hardware can have the same specs yet sound quite different?”]
Sorry, back on topic: so, if you train a listener, either actively or passively by merely “consuming” HRA content, to hear the improvements that true hi rez content‡ encodes, then your average Joe or Jane can in fact hear the improvements and statistically prefers a higher fidelity version. [Hint: there’s more “there” to hear and enjoy] I am in my mid–sixties, and I can distinguish between an HRA master and a Red Book child generated from that master. Am I special? F•ck no, I have simply trained my hearing mechanism to notice such things.
Bottom lines;
• Even if you can’t hear it, higher resolution can carry subjectively higher fidelity if all the 10,000 choices, large and small, made during a production are all aimed at that goal of higher sound quality.
• Sound quality and quality of performance/musicianship are unrelated. On those rare occasions when both coincide, you have…magic.
‡ Another significant aspect of this discussion is folks who get all in a huff about “$12,000 power cords” and other audiophilia. Dude, just because you can’t afford a piece of gear, it doesn’t mean it’s invalid. Yes, of course there are vendors in the CE audio world who are making beaucoup bucks off of endodontists and investment bankers but, sorry to be blunt; when was the last time anyone listened to a state of the art, $100k mastering playback rig or, in the same vein, any playback system that cost more than $8k (an arbitrary sum that usually entrains better fidelity)? Did you “hear things I never heard before”? Most likely. I don’t mean to be snarky or condescending but, within limits, you really do get what you pay for.
Since analog and plain ol’ physics underpins all “digital” electronics, then pretty much every aspect of a recording and playback system matters if you want to preserve fidelity from performer to consumer. That USB printer cable you might grab to plug in your audio interface? In terms of fidelity, it’s worth every penny, or about $1.99. “Digital” is not a panacea, and neither is a higher bit rate or longer word length.
MI and most pro audio gear is built to a price point, not to a performance spec. Yes, there are exceptions, like a Maselec or Manley EQ, a Studer open reel deck or a Milab or DPA microphone. They cost more cuz higher quality parts and more careful engineering result in a clearer picture of a performance. As a kid, I spent my time as a staff engineer in a giant recording complex (Criteria in Miami). We used MCI desks and tape machines exclusively because the owner was a high school buddy of MCI’s founder, and it wasn’t until much later that I got to experience the higher fidelity that better quality and more costly gear can bring to the party. That said, that studio’s gear could be thought of as one giant composite effects pedal, placing an undeniable sonic stamp on all the music that was made there.
I think I know what you mean here but a good musician should change any of the many parameters under his/her control to adapt to different situations. For example if a recording situation lacks subtlety or definition we might paint with broad strokes and bold colors so as to not make that weakness obvious.
Another example of this was mentioned in another thread recently. Bluegrass musicians often use one microphone and position themselves around it to create a ‘mix’. They trade places during the performance to adjust levels and vary the timbre of their instruments to give the microphone sounds it can best use.
I reckon you been reading this thread and came across the 3 exact versions of a song from 3 different rates: 48/24, 192/24 and192/32.
They were all exported from a 48/24 tracking.
In your pro opinion, would there be a quality difference even though two of the copies were upsampled?
I hear a difference, though i might be fooling myself by concentrating to much on the numbers and not what i hear per se. But i do hear somethings being different sounding. Like a bit more 3 dimensional.
Is it in my head bro?
Lol.
sorry to side step on this, but if you can truly tell the difference in a blind test, then it is not in your head. My hearing is fairly good. I can hear well in the 17k range and I can pick up reverb tails and early reflections well. In a test I ran with the help of a friend in a very high end orchestra hall, I could certainly tell the difference in ambiance in a 24bit recording vs 16 bit in a blind test repeatedly. Bit depths are easier to pick up by ear than sample rates.This was done with two same mics recording the same cellist simultaneously. I no longer have those recordings as they were damaged or I would post them here.
If you are hearing differences in upsampled recordings, you could be picking up the artifacts subconsciously. It is hard but if you do have super hearing, you can likely hear these artifacts. But they are usually not easy to pick off in a blind test.
Thanks for the video! quite informative.
The issue why this level of blind testing cant be done on a large scale is that the individuals who can truly tell the differences are already a statistical anomaly. They are rare. We are talking 0.01-0.05% of the population. So any or all statistical analysis of hearing tests will be inconclusive unless we round up a decent enough sample size of people with perfect pitch or heightened listening and perform the test in a controlled environment with ideal samples.
The way I generally do a blind test is with help of a friend and 3 second pauses between switches. Sometimes my friend switches the samples and sometimes plays the same sample again to trick me! 
Hey man,
Yes, there should be a difference, because they were upsampled! As I mentioned SRC’ing, implying downsampling to arrive at a deliverable, upsampling is also SRC’ing except in the other direction. In the case of upsampling, as with all sample rate conversion processes, a large contributor to the overall sound is the crucial windowing and subsequent filtering of the resulting samples. With upsampling, you really are listening to their windowing and two filter behaviors above all else since there is no new information. An FFT of an upsampled recording reveals the telltale passband of the original sample rate; there’s no audio above Nyquist.
I mentioned two filters; one the SRC process’ filter, the other is your DAC’s reconstruction or anti–image filter. With digital audio, you’re always listening through the DAC’s sonic signature. If you hear a “bit more (of a) 3 dimensional” aspect, then that’s your DAC performing better with more samples to work with or a higher data rate; two side of the same coin. Another aspect of DAC performance that comes in to play w/this scenario is jitter performance. Lower jitter usually resulting in better soundstaging and more solid bass, so you may be hearing better imaging because of that.
Factoid: el cheapo CE DACs and commodity audiophile playback software typically offer upsampling as an option, for those very reasons. It’s not accurate, but subjectively it can sound “better” to some. BTW, I have an R-2R DAC in dah house destined for a future review, and it offers both non–oversampling and oversampling modes via a front panel switch. I don’t like the oversampling mode myself, as it provides more crispness than an air fryer, but some folks might like that if their rig/room/hearing is a touch dark.
Hey FluteCafe,
Thanks for your anecdote, you’re lucky you got to do that comparison. I once had a chance to compare a jazz combo, captured live in a studio to both hirez PCM and DSD1. It was instructive!
Your comment about test recordings jogged my memory…Morten Lindberg, owner of record label 2L and audio engineer/producer extraordinaire, used to maintain his “test bench.” It’s gone from their current site but lives on thanks to the generosity of the Internet Archive, may the Gods bless them. See:
Morten now records only at DXD rates, and uses high quality SRC. So, rather than listening to fake high rez, try truly HRA audio that has been carefully downrez’d and/or transcoded to DSD and/or MQA…a much more meaningful test. I know, most of you don’t have full or even partial MQA decoding but, as my grandpappy would have said if asked, “Word. If you ain’t listened with your own two ears, you can’t have an opinion.” Same goes for DSD; the test bench has DSD1, 2 3 and even DSD4 versions of many of the clips listed. While DSD1, 2 & 3 are roughly equivalent to 88.2, 176.4 & 352.8, DSD4, a.k.a. DSD256, can be thought of as upsampling as well as transcoding. Very few pro DACs “do” DSD (Tascam, Merging, DAD) and the majority of CE DACs do either LPCM or DSD well, but not both. So, exercise you ears and gear, and take a listen.
I’m totally in the “I can’t tell the difference” camp. However, a few things have sort of crept in that I noticed but never gave any thought to with 96.
They played over half the song and screwed up bad. They decided to try again. I changed back to 24/48. When I got back to my studio, there WAS a difference between the two recordings. I can’t even tell you what it was. But the higher resolution just seemed like everything was completely audible at all times. The 24/48 version seemed like everything on the whole was sort of glued together where things just didn’t stand out as much.
I’ve always chalked this up as real instruments playing a role in the sonic field. What I mean is, when I’ve done tests years ago, anything distorted or even direct just didn’t seem to make a difference with different sample rates to me. This was the first “woah” moment for me and let me tell you, I’ve done some really intense testing on this but not much by way of orchestra instruments.
When all was said and done, this turned out to he one of the best projects I had ever done. By this time, I had totally forgotten about the 24/96 thing. The guy was so happy with what I had done, he came back in to record 4 more songs. I had made a template for his last project that we used pretty much verbatim.
This time though, I was back at 24/48 and did all 4 songs that way since we didn’t need any guide tracks or performance tracks. When I was done with these 4 songs, I put the 5 I did at 24/96 and these new 4 at 24/48 on an audio cd to listen to the whole project.
I wasn’t even thinking about the sample rates as I had forgotten about them. As I listened to the first 5 and then the last 4, I noticed I was hearing that same thing I heard with the school band. The 5 at 24/96 just had this “thing” going on that allowed each instrument to shine very audibly without sticking out like a sore thumb as “that’s just too loud”.
The 24 /48 mixes were still as good as I could have gotten them, but they had this cohesiveness that though good, sounded like the instrument identities weren’t as apparent. I used all the same settings and we did not hyper compress anything. To this day it sort of bothers me and makes me want to commit to 24/96. I keep telling myself it’s overkill but I’m hearing “something” that I like better.
One more interesting thing. There was a guy that used to be here on the old site. He went by Tony. He was an engineer that sent me loads of mastering work and I even played guitar on a lot of his client work. I happened to listen to some stuff I had done with him and noticed this same thing with the instrument identities just shining a little more. Then I remembered, all his recordings were 24/88.
I’ll never admit I can hear a difference, but I hear “something” that is enticing me to go to 96 just to see if it’s consistent. At the end of the day, I just want great recordings that I and my clients love. I think I’m there at 24/48 but that “something” at 96 (if it really exists) would be welcome in my realm. Lol!
Danny