I just went and tested my hearing. I'm very surprised to say I can hear right up to 22.5 Khz ( as high as the tests I tried went ) although I have some minor hearing loss in my right ear, which I never realized until I started using bluetooth with my cell phone and switched ears. I'll be 30 this august, and I've played guitar for 15 years as well as attending over 100 concerts. I used to listen to headphones at an unhealthy volume but now I enjoy music at safer levels with the acception of some indoor concerts/shows.

Quote:



That's interesting. Did you test your hearing yourself or did you have an audiologist test you?

Quote:



I took 5 different tests online. I don't really think it's necessary to pay an audiologist to find out what frequencies I can hear, although doing a complete examination with an audiologist would of course yield more in-depth results. I'm not at all sensitive to such high frequencies, but I can hear the tone.

Quote:



At what sampling rate did you play the tones at? You're probably hearing artifacts and not the actual tone.

There could be other variables at play too. I'm not an expert at hearing tests, but it was my understanding that there is a procedure to follow to set the volume of the tones. And, yeah, there could be artifacts that are being heard.

Not trying to say you can't hear up to 22.5K, but it is so unlikely given your age and the loud volumes you used to (and still do at times) listen at that it's reasonable to explore whether there is some other explanation for your conclusion you can really hear tones at that frequency.

Quote:



If your audio card is subject to aliasing (as many/most are) then you could try the test 50 times and the result would be the same. In any event it is unlikely in the extreme that you can hear 22.5 kHz... I would try using some different playback hardware to see if you get the same results.

I'm not using my audio card, I'm using spdif out and the rig in my sig. I would say the tones I could hear were obvious until 21Khz, and I had to jack the volume beyond there. It shouldn't really matter what volume a frequency is played as you can hear it or you can't, but I'm not doubting other things could be affecting the result. I know I can hear better than average at as I've never met a dog whistle I couldn't hear, and I'm often annoyed by high freq given off by electronics ect when other people are unaware. My little sister has those annoying "mosquito" ringtones that she uses at school so teachers can't hear her phone ring, and I can hear those also. Another semi-unrelated thing I've noticed is that I'm always aware of what song/artist is playing in the background in public places when other people I'm with aren't even aware there is music playing (such as over a PA in a store/mall) at all. And no, I'm not schizophrenic . Anyway, if one of you guys wants to send me all the rules I need to abide by to play the frequency game I'll give it a try.

Quote:



I'm not going to say that it is impossible for you to be able to hear 22kHz at thirty years of age but at a guess I would say that if you really can hear 22kHz you are a member of a tiny group of people on the planet.

BTW, it will make a difference the volume that the frequency is played back at. Human hearing is not linear with a sudden cut off at a particular frequency. Human hearing rolls-off gradually from around 12kHz, meaning the higher the frequency beyond about 12kHz, the quieter it will appear. The same is true for low frequencies below about 200Hz. Therefore, the louder the volume, the more likely you are to hear higher frequencies. An Audiologist will run tests at fixed levels to avoid this issue.

A word of warning though, a colleague of mine was running a frequency test with a class of degree students. He was certain that some of the teenagers would be able to hear 18kHz so he kept gradually turning up the volume. Eventually one of the students was able to perceive it, when he saw smoke coming out of the tweeters! On closer inspection, the tweeters on the Blue Sky monitors had completely melted!!

G

Gregorio, I know this thread is focused on bit depth, and on that subject when I've listened to the same recording at 16 vs 24 bit depth I could hear no difference and I tend to agree with the presentation of your arguments.

However, that same recording (which is 24/96) that I got off of hdtracks.net does sound remarkably better than anything I've ever heard off of a redbook CD before. I'm sure it mostly has to do with the fact that it is just a very well done mastering. Now I've been reading through the thread and I'd like to hear your thoughts a bit more defined on the effects of higher sample rates and their ability to affect the quality of the playback/recording.

It just seems to me, from a purely laymen point of view, that if you take an analog sine wave and sample it 96 thousand times a second vs 44 thousand times a second the end result is going to be a smoother, more accurate, digital reprsentation of that sine wave. The thing that struck me most about these high quality recordings was how smooth everything sounded. Especially things like the minute vibrations and decay of a cymbal shimmering after a drum stick hit it. The sound tapered off in an extremely realistic way that I've never heard in a recording before. Also, the overtones of stringed instruments in classical music seemed to sound much more pronounced and accurate.

These types of things seem like they might benefit from having higher sample rates, reproducing the analog wave more accurately. So, bit depth aside, are high quality recordings worth it for the sample rate increase alone? Or are all these DVD-A/SACD/DSD recordings just well mastered music that could have been put on a CD and no one would be the wiser (assuming stereo, I know multichannel stuff wouldn't fit)?

Smoother waves, but doesn't resampling it add distortion and maybe even phase differences? The resampling schemes alone vary so greatly and noticeably sound different to me.

Quote:



As far as digital audio theory is concerned there is a problem with the first statement in second paragraph I've quoted. If you look at a graphical representation of the digital data in a digital audio software package, waveforms recorded with higher sample rates (and bit depths) it will appear to look smoother and more like an analogue waveform on the screen. This is useful when editing digital audio but is rather mis-leading unless you fully appreciate that you are looking at a graphical representation of the digital audio data rather than what will actually come out of a DAC once that digital audio data is converted back to analogue. The "rule" of digital audio theory is that providing the sample rate is at least double the highest audio frequency then the waveform can be reproduced perfectly. This means that recording say a 12kHz sine wave at 96kFs/s or even 192kFs/s is not going to reproduce the 12kHz sine wave any more perfectly than the already perfect 44.1kFs/s recording.

I still haven't had chance to recalibrate my work system to actually analyse the linked 24/96 track. However, I don't doubt that those of you who heard a difference actually did hear a difference. I too have heard differences between some recordings released in both 24/96 and CD format. Unfortunately there are many potential reasons for this.

The first thing I would say is that if you can actually hear more pronounced overtones and smoothness to cymbal decays (for example) then by definition, what you are hearing is within the audible frequency spectrum and therefore just as capable of perfect reproduction at 44.1kFs/s as at 96kFs/s. However, there maybe other factors at play, here is an example. Let's say that a particular cymbal is very exposed in a track and that it has a sonic characteristic which includes an obvious peak at around 16kHz. During the final phase of mastering or mix down noise-shaped dither is applied and it just so happens that the algorithm used redistributes the dither noise centred around 16kHz. It's entirely possible that this dither noise could to an extent mask the lovely cymbal decay and be easily missed by the mastering engineer, especially if it is played back on audiophile headphones which often enhance the higher frequencies. Another possibility is that the reconstruction filters in a particular DAC could be causing more noticable artefacts at 44.1kFs/s than at 96kFs/s. Although this effect is very small and most likely to be noticed with cheaper convertion processors. Mostly though I would expect differences between versions to be a direct result of some mastering process.

It should also be noted that I have noticed that it's not uncommon for 16bit versions of commercially released hi-res recordings to be either deliberately of lower quality or end up being of lower quality because no one puts any real effort into making it sound as good. If you think about it, a record company (and consumers) wouldn't be too happy if they sold a recording at a premium because it's in a Hi-rez format if the 16bit version had the same or better sound quality. The whole industry, from record labels to consumer equipment manufacturers is poised to take advantage of the hi-res formats and the equipment capable of handling it. The whole marketing edifice would fall flat on it's face if consumers actually knew that hi-res was no better than good old CD.

I might have a little time tomorrow to download the file linked in an earlier post and have a listen for myself. I might even have a go at creating a 16bit version of my own for comparison by head-fiers.

G

Quote:



Just as an FYI, the 24/96 tracks that I downloaded and am referring to were from hdtracks.net, which is offering a free "sample download" of some of the stuff they are selling. I have not listened to the linked music earlier in the thread.

I should also note that I wasn't comparing these to 16/44 versions of the same music, of which I don't have access to. I was just remarking at how amazing the reproduced audio sounded in general.

Now, as to the part of your post that I quoted. I do understand what you're saying regarding the sample rate having to be double the highest frequency for a reproduction of the sound to be possible, but I'm still not seeing how that is related to my question about the QUALITY of the reproduced sound. I'm sure it is and I hope you'll help me understand.

In my mind I picture an analog sine wave on a graph. The X axis of the graph represents time. If I'm trying to digitally do a "connect the dots" game on that sine wave, it would obviously seem beneficial to have more dots. The more dots I have the smoother my wave form is going to be. This here is the crux of my question. Frequency response aside, which I understand a 44Fs/s sample rate allows me to go up to 22khz in audio, wouldn't the smoother wave form of a "hi res" recording still be a more accurate reproduction of the sound than the less smooth (44) version?

Quote:



Your error is that in your mind you're "connecting the dots" with straight lines. The correct lines are bandlimited interpolations, which are how real oversampling DACs work. When you apply bandlimited interpolation between the points, you will find that adding more points by increasing the sample rate results in no additional quality gain. The new points will fall exactly on the line that you originally drew. Therefore, no additional accuracy.

This link has been dug up thousands of times before, but it looks like you haven't seen this before, so:

http://www.lavryengineering.com/docu...ing_Theory.pdf

gregorio, do you hear the differences between 32 bit and 16 bit?

Quote:



/facepalm

If it's already extremely difficult to tell the difference between the extremely low noise floors of 16 bit and 24 bit, why would the even lower noise floor of 32 bit be any easier to discern?