24/96 and 24/192 is no good. MP3 sounds better!

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
This report makes what I said true then.
It doesn´t matter if you hear it or not, the brain though can feel the difference.
It´s like very low bass. You feel more on your body than you hear it.
The are parts of the human brain we don´t know how they work.
I think the ultrasonic frequencies from musical organs can be senced from a human and that´s what makes the difference between real life music and hi fi that is restricted to 20KHz bandwidth.
 
BTW - just for reference, NTSC h-sync frequency is 15.7 kHz... not exactly ultrasonic, but many people can't hear this, and so it could perhaps be considered ultrasonic for those individuals. I have no idea what frequency those motion sensors operate on, but I suspect it's well over 20kHz.

Also, redbook CD does preserve phase almost perfectly, right up over 10kHz. Even towards 20kHz, the phase is maintained very well. I personally have my doubts about the amplitude and phase accuracy of PCM sampling systems as the Nyquist frequency is approached, but that is another matter, and for what it's worth I also have serious doubts about the effectiveness with real music of SACD's DSD sampling method above about 10kHz or so. DSD will *drastically* alter the ultrasonic content of transient events, since there is insufficient bitrate to resolve these events. High rate PCM on the other hand will maintain accurate phase and amplitude response right up to 48kHz (for 24/96) or even 96kHz (for 24/192, and will accurately capture transient events all the way up into the ultrasonic region.
 
sam9

If seems to me that if you havr for instance a 50khz fundamental that induces a harmonic in the audible range, say 12.5khz, once the 12 12.5 hz harmonin is captured (I.e., recorded) it matters not one whit if the the subsequent signal chain includes the 50khz fundamental. When played back with technology having an upper limit of 20kHz to humans with a simliar upper limit, they will hear the 12.5kHz harmonic even though the fundamental is no longer present.

To this extent, what happens above 20khz is important because distortions in the audible range can be introduced from tones outside the audible range (this is why amp designers like like their designs to remain linear out to 50kHz) or because the original source contains inaudible elements that generaste fundamentals at a lower frequency.

The gamelon experiment is interesting but the conclusion is flawed in a fairly obvious way. Induced harmonics don't just occur in electronic signals but in airborne sounds as well. A missing piece of the puzzel is whether or not the experimental set up created acoustic (rather than electronic) audible artifacts that would be present in the original live source. The descrition of the experimental set suggests to me that to be a very real possability.
 
Most of my CDs have some frequencies up to 22 kHz (according to DSP spectrum analyzer for Winamp), and I've recorded the analog output of my Yamaha CDX860 CD player (450 € in 1991) : The loss at 22 kHz is no more than 1 or 2 db.
Everyone says CDs are lowpassed at 20 kHz, and according to the spectrum analyzer it's obviously not the case. Everyone also says that CD players are lowpassed at 20 kHz, and it's not the case for mine. Is it really an ultrasonic monster, or is it just a rumour ?

This said, if ultrasounds can really affect human hearing, the effect is far less than in electronic gear, and while it is very easy to hear those spurious tones when they are produced by an ampli or a speaker, I've never heard of any experiment that could make them audible otherwise.

We ran the test in this thread
http://www.musicplayer.com/ubb/ultimatebb.php?ubb=get_topic;f=3;t=000822#000000
and I just repeated it before writing this.

Generate two waves : a sine 2 kHz below your hearing cutoff, the other 2 kHz above. For me, it is 14 kHz and 18 kHz.
Make two files :
One with the 14 kHz on the left channel, and the 18 kHz on the right, the other with both frequencies mixed in mono.

The result is always the same, at low level, I can only hear the 14 kHz.
At high level I can hear the 4 kHz intermodulation in the mono file only. I can push the volume much higher on the stereo file without hearing anything else than the 14 kHz.

I think it's better not playing these tones for more than one or two seconds at high level, I've heard it can fry the tweeters.


Conclusion : the presence of intermodulation distortion is a defect of the gear. It doesn't occur when two separate amplis and speakers (here, left and right channels) play the frequencies separately instead of together in one (mono) channel.


Setup :
Waveform generated by SoundForge 4.5 at 44.1 kHz
Playback : Marian Marc 2 digital output (no resampling)
Sony DTC 55ES as external DAC
Arcam Diva A85 ampli
Dynaudio Gemini speakers at 1.2 meters right in the axis of the tweeter. I was exactly between the speakers, each one turned towards my ears.
I then realized that my head could mask each other channels at those frequencies, so I moved back 2 meters. I could still hear the 14 kHz in both ears. But there was still no intermodulation. I also tried facing a speaker (thus with the other in my back). No more result.

The stereo file was played 10 db louder than the mono file without showing the intermodulation that was in the mono file.

Guessing the actual level playing a white noise of same peak level instead of the sines, I think the playback was about 95 and 105 db in the room. More or less 10 db (I'm not good at guessing levels in DB, It would have been easier with a pink noise, but I couldn't find any in my backups :) ).
 
hifiZen said:

...Symmetric FIR filters have perfectly linear phase, and thus produce precisely zero phase shift at all frequencies.
Could be me, but isn't this a contradictio in terminis?

As I understood it, the phase shift is linear over frequency, which is perceptually quite natural. However, (anti-)symmetric FIRs are not zerophase filters, and will therefore introduce phase 'distortion'. Using a noncausal filtering operation with a symmetric FIR can give zero phase shift.

Please correct me if this view of linear phase response is incorrect..
 
Hi petracci

I think hifiZen meant the right thing but was a little unexact with his description.

Symmetric FIR filters introduce a constant delay which will not give any phase-distortion at all but a phaseshift that is proportional to frequnency.
It's effect is the same as a short delay line (apart from the amplidude response of course) or the effect of pushing the start button of your CD player some fractions of a millisecond later.

This phase behaviour is the reason why some people think of FIR filters being the best solution for active loudspeaker crossovers.

regards

Charles
 
Hi Charles,

phase_accurate said:

Symmetric FIR filters introduce a constant delay which will not give any phase-distortion at all but a phaseshift that is proportional to frequnency.
The phaseshift is then linearly proportional to frequency, right? But it is different from the original unfiltered signal, so you could still call it a phase distortion. But probably this refers to nonlinear phase differences?


This phase behaviour is the reason why some people think of FIR filters being the best solution for active loudspeaker crossovers.

I recall reading that minimum phase filters could be just as good (if not better) since they are closest to zerophase filters. Any comments/info on that?

Thanx,

Petracci
 
Hi Petracci

This is definitely not phase distortion. If you move back a little from your loudspeakers this might cause some degrees more phase lag to a low frequency sound but several wavelenghts for a higher frequency tone. But the temporal alignment of fundamentals and their harmonics will not be altered.
A constant delay causes no phase distortion (the correct term would be group delay distortion anyway).

The crossover question is off-topic here so I make it short. There are the following possibilities (at least to my current knowledge) to build crossovers that don't introduce phase distortion (i.e. that are phase accurate or transient perfect :) ) each of them having their inherent advantages and disadvantages:

1. first order crossovers
2. FIR crossovers
3. filler driver crossovers
4. subtractive crossovers (see Nelson Pass' webpage)


Any higher order conventional crossover will definitely introduce phase distortion. This is taking the crossover only into account. The added response of the drivers will make things worse.

regards

Charles


P.S.: dorkus' post arrived while I was typing. I can fully agree with it.
 

is there such thing as a zero phase filter
[\QUOTE][\B]

It's noncausal, forward/backward filtering with a symmetric filter, so no realtime implementations here. In the frequency domain, only the magnitudes are changed compared to the original.


...some degrees more phase lag to a low frequency sound but several wavelenghts for a higher frequency tone.
But the temporal alignment of fundamentals and their harmonics will not be altered. A constant delay causes no phase distortion
[\QUOTE]

So if you correct the frequency lags for full wavelengths and you compare the phases of frequency components of the orginal signal and the filtered signal, they will be the same?
 
In order to post a little more on-topic regarding this tread here's my two cents:

In my opinion MP3 and the like was never intended for serious listening but as a new technology for portables.

As far as the often stated "CD quality" goes, my opinion is that the CD doesn't make full use of many capabilities of the human hearing. I think when the standard was born (more than 20 years ago) it was with the goal in mind to be able to implement it with reasonable cost (in the end you want to be able to SELL what you invent).
In the beginning there were even CD-players using only 14 bit DACs simply for economic reasons and those were still quite expensive !!!
The dynamic range of the human hearing is far better than what the CD's theoretical 96 dB allows !
When the first players and CDs arrived people were quite astonished by their ease of use and the apparant absence of noise (one of the main sales arguments back then) which aren't any guarantee for high quality sound on their own. I think everybody on this forum has had the opportunity to listen to old noisy recordings that beat many modern recordings in respect to naturalness and accuracy.
But it IS possible to produce good sounding CDs. I think the bad sound of many CDs has nothing to do with the technology itself but with the people who produce them (i.e. musicians, sound engineers,producers,.....) and aren't able to use the potential of this medium to the full extent.

But after 20 years I think it is time to introduce new standards that are getting closer to our hearing's capabilities and that use the offerings of today's technology.
I have to admit that I haven't listened to DVD-A so far but have been able to hear the differences between SACD and CD (from hybrid discs to have a fair comparison) and it was definitely a step forward.

As far as analog goes: I don't listen to vinyl currently. But I own some records that definitely sounded better on my old TT than the CD version. And my TT was far from being high-end. It was a very old German-made Dual with a Shure M95 cartridge.


Regards

Charles
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.