It isn't actually possible to use a transform on any signal and obtain a perfect semblance of the original. You always introduce error regardless of the transforms you are using. Now, you can reasonably argue that at some level it doesn't matter -- that electronics components nor human ears are perfect and within that error we have a great deal of breathing room. But, no, you cannot create a digital signal that is an exact replica of an analog signal -- there aren't enough atoms in the universe to encode onto.
I have a big space to fill. Any reason why I should not go for two of these baby's and wire them up together? Is it as easy as splitting the audio signal, or would that be a no no to the audiophile?
But, no, you cannot create a digital signal that is an exact replica of an analog signal -- there aren't enough atoms in the universe to encode onto.
Whilst this is kind of true, it doesn't mean that vinyl has more information than CD.
CD does two things that vinyl doesn't: it samples the signal in time, and it quantises it in amplitude.
The time sampling limits the time resolution, and therefore gives an upper frequency limit. As I pointed out, the upper frequency limit is the same for CD as it is for vinyl, except any given vinyl record will decrease in bandwidth each time it is played.
If you take an analogue signal, and sample it at anything over twice the maximum frequency component, you can reconstruct that signal perfectly. This is a simple application of mathematics, it is a fact. Note, I have not said anything about quantising the signal.
Next, something a lot of people don't understand is that quantising is exactly the same as adding noise to the signal. But there is noise in a vinyl system, too. CD quantisation adds less noise to the signal than the vinyl recording and playback system. Noise is not information.
So, if you take an analogue signal, it will contain noise. There is absolutely no point in using more bits, just so you can accurately encode the noise, because you would need a perfectly noiseless playback system (impossible to achieve) in order to play back exactly what you recorded, and this would be pointless anyway, because the noise is, well, noise. It is not the signal.
I have a big space to fill. Any reason why I should not go for two of these baby's and wire them up together? Is it as easy as splitting the audio signal, or would that be a no no to the audiophile?
Thanks!
All the reports that I've read so far indicate that the iPod HiFi is well capable of filling a large space with sound. I don't think there's any way to wire two together (at least not if you want to source music from an iPod docked into the top of one of the HiFis).
The iPod HiFi does have an external input, however. So, you could make up a cable that had a stereo connector at one end, and four connectors at the other: right signal x 2 and left signal x 2. Connect the two right signals to one iPod HiFi, and the two left signals to the other iPod HiFi. Then all you have to do is synchronise the volume. The only problem is, you can only do that by ear, because the HiFi doesn't have a volume read-out. Also, I don't know if the HiFi works without an iPod docked into it.
I have a big space (studio). I was looking to splurge on a set of these [ http://www.genelec.com/products/bi_amp.php ] some time this year.. Now I'm heading Apple's way again.
It better work without the ipod. I am still a happy owner over ipod v1.0 with firewire connector! Only apple product I felt safe to early adopt. HiFi will be the second.. I'd be driving the music to it via airport. I'll start with one and see how it goes. Thanks!
I have a big space (studio). I was looking to splurge on a set of these [ http://www.genelec.com/products/bi_amp.php ] some time this year.. Now I'm heading Apple's way again.
It better work without the ipod. I am still a happy owner over ipod v1.0 with firewire connector! Only apple product I felt safe to early adopt. HiFi will be the second.. I'd be driving the music to it via airport. I'll start with one and see how it goes. Thanks!
These are probably very good speakers, but honestly they're
looking b**t-u*ly. What about pricing? (too lazy to google)
These are probably very good speakers, but honestly they're
looking b**t-u*ly. What about pricing? (too lazy to google)
HAHA! To each his own. They look ok in da flesh... or should I say cast aluminum. The weigh 15KG each. Sounded A M A Z I N G in the shop studio. Require a sub-woofer for big spaces. They have built in amplification. I think they were around 350 euro ($400) each. So, not cheap, but not crazy audiophile pricing. They are worth checking for sure. I'm going for a HiFi and will see how far that gets me.
HAHA! To each his own. They look ok in da flesh... or should I say cast aluminum. The weigh 15KG each. Sounded A M A Z I N G in the shop studio. Require a sub-woofer for big spaces. They have built in amplification. I think they were around 350 euro ($400) each. So, not cheap, but not crazy audiophile pricing. They are worth checking for sure. I'm going for a HiFi and will see how far that gets me.
Well, no offend here As i stated somewhere above,
Genelec is a brand you have to factor in, IF you DO WANT
and NEED extraordinary quality regarding sound. No question here.
But, (and this is strictly a personal point of view, and hence a matter
of my very own taste) i do prefer Tannoy in look and sound over
Genelec (and other comparable brands in this price range).
sic and cheers Though the genelec you propose seems
If you take an analogue signal, and sample it at anything over twice the maximum frequency component, you can reconstruct that signal perfectly.
Nyquist's law always seemed kind of like a guess to me in school.
If it is really true, then why does 24/96 digital audio sound so much better than 16/44.1? Why does SACD sound even better still?
I think that people that get caught up in the math of sound end up missing out on the good stuff. You just can't get a good sounding stereo if you rely on specs and math - CD sound is harsh, brittle and dull sounding, and so is solid state sound.
But, no, you cannot create a digital signal that is an exact replica of an analog signal -- there aren't enough atoms in the universe to encode onto.
You don't need to create a digital signal that is an exact replica. You just need the signals max frequency to be less than 1/2 the sampling frequency, and you can regenerate a perfect copy of the digitized analog source signal if you have an ideal-low-pass filter. Now, ideal-low-pass filters are hard to come by. That's why you oversample and use a good FIR filter within the digital domain.
I don't mean to pull rank, but do have a degree in electrical engineering, during which time I focused on signal processing, and after which time I started working in the RFID industry. I do know what I'm talking about.
Quote:
Nyquist's law always seemed kind of like a guess to me in school.
Then you didn't pay attention. Nyquist's law can be proven mathematically (relatively easily). Note that it has been dubbed a "law," which suggests that it's not some kind of hack. Moving on, if our mathematics are wrong, then either side of an argument on mathematical principles is flawed, and you should give up.
Quote:
I think that people that get caught up in the math of sound end up missing out on the good stuff. You just can't get a good sounding stereo if you rely on specs and math - CD sound is harsh, brittle and dull sounding, and so is solid state sound.
This is a dumb argument. If a digital signal can be reconstructed exactly into the source signal, then it will sound exactly the same as the source signal. The reason why it doesn't is because the system you're using doesn't do a good job of reconstructing. If it could, you wouldn't know the difference.
Incidentally, there's also plenty of scientific evidence suggesting the placebo effect is real. I think that's really the culprit here.
Nyquist's law always seemed kind of like a guess to me in school.
No, it wasn't a guess. If you take any signal, and sample it at a rate at least double the highest frequency that signal contains, you have captured it perfectly. Take a signal x(t), and sample it in time, to give x[n]. There is one, and only one, signal that can be represented by x[n], and that's x(t). Therefore, it is possible to perfectly reconstruct x(t) from x[n]. (Yes, I know this isn't a proof, but it is the general idea and principle).
Quote:
Originally posted by e1618978
If it is really true, then why does 24/96 digital audio sound so much better than 16/44.1? Why does SACD sound even better still?
See my previous explanation. Also, placebo is a distinct possibility.
Quote:
Originally posted by e1618978
CD sound is harsh, brittle and dull sounding, and so is solid state sound.
listen, I'm not trying to tell you that CD or solid state electronics sound better. You may prefer vinyl and tubes, fair enough. But don't kid yourself that it's because such a system is giving you "more of the original sound"; it is not. It is giving you less, and what it is giving you is more distorted and compressed (in terms of dynamic range). It is these distortions and dynamic range compressions that give vinyl and tubes their warmer, less harsh sound. If you want to hear exactly what was recorded, unfortunately, that's impossible, but the closest thing you can get is a recording and playback system as I outlined a few posts ago.
You don't need to create a digital signal that is an exact replica.
CD players, even cheap ones, can sound dramatically different from each other. Not only that, but CDs made now sound a lot better than CDs made in the early 80s.
Even with lossless compression, the iPod mini sounds really bad when you hook it up to a stereo, much worse than any CD player.
I don't think that your "exact replica" is as exact as you think it is.
I used to hear well beyond 20 -20,000 but it was a while since my last test.
I very much doubt that claim, as even newly born infants are hard pressed to reach those limits. What i'd gather you are actually 'hearing' are harmonics of the higher/lower frequencies. Second, third and perhaps fourth order harmonics, not the actually figure over or under the numbers stated.
I very much doubt that claim, as even newly born infants are hard pressed to reach those limits. What i'd gather you are actually 'hearing' are harmonics of the higher/lower frequencies. Second, third and perhaps fourth order harmonics, not the actually figure over or under the numbers stated.
It depends what he means by "well above". With hearing, just like anything else, there is a spread of capabilities within the population. A very, very small number of people really can hear beyond 20 kHz, just like a very very small number of people have an IQ of 170 or more, and a very very small number of people can run 100 m in less than 9 seconds.
Also, I don't know what you're talking about with the harmonics thing. I think it is possible that when he tested his hearing, and was driving a speaker with greater than 20 kHz, the system could have produced a tone below 20 kHz (i.e. a sub-harmonic) it really depends on the distortion mechanisms in the sytem he was using.
You don't need to create a digital signal that is an exact replica. You just need the signals max frequency to be less than 1/2 the sampling frequency, and you can regenerate a perfect copy of the digitized analog source signal if you have an ideal-low-pass filter. Now, ideal-low-pass filters are hard to come by. That's why you oversample and use a good FIR filter within the digital domain.
I don't mean to pull rank, but do have a degree in electrical engineering, during which time I focused on signal processing, and after which time I started working in the RFID industry. I do know what I'm talking about.
Great! And I don't mean to pull rank on you, I use analog to digital transforms and every day as well and I am getting my third degree!
Now sit down.
I was simply pointing out that you are making approximations about what signals are relevant and those that aren't. You can claim the ability to obtain a good enough replica because you are limiting the frequency space you are interested in to the range of human hearing, which isn't exactly the same as all frequencies.
Great! And I don't mean to pull rank on you, I use analog to digital transforms and every day as well and I am getting my third degree!
Now sit down.
I was simply pointing out that you are making approximations about what signals are relevant and those that aren't. You can claim the ability to obtain a good enough replica because you are limiting the frequency space you are interested in to the range of human hearing, which isn't exactly the same as all frequencies.
Really? (about your qualifications), well this is strange post for you to make, then.
We are talking about vinyl vs. CD. They both have the same bandwidth (as long as you only play your record once). Therefore CD is losing nothing to vinyl in frequency terms.
Quote:
Originally posted by hardeeharhar
This fact is why Nyquist's Theorem isn't a law.
?? The fact that you can't capture a signal with infinite frequency content by sampling it has got bugger all to do with whether Nyquist's Theorem is true or not. In fact, it is Nyquist's Theorem that states that you cannot capture a signal with infinite frequency content by sampling it.
We are talking about vinyl vs. CD. They both have the same bandwidth (as long as you only play your record once). Therefore CD is losing nothing to vinyl in frequency terms.
I am not talking about vinyl... I was trying to get splinemodel to be more precise in his language, that is all...
?? The fact that you can't capture a signal with infinite frequency content by sampling it has got bugger all to do with whether Nyquist's Theorem is true or not. In fact, it is Nyquist's Theorem that states that you cannot capture a signal with infinite frequency content by sampling it.
It isn't a question of whether it is true or not, it isn't a law because it doesn't cover all frequency space..
It isn't a question of whether it is true or not, it isn't a law because it doesn't cover all frequency space..
Well, I suppose that depends. Does infinite frequency space exist? And if it does, just sample it at infinity x 2 (yes, you can have different size infinities e.g. the number of irrational numbers >> the number of integers, but they are both infinite in number).
Comments
It isn't actually possible to use a transform on any signal and obtain a perfect semblance of the original. You always introduce error regardless of the transforms you are using. Now, you can reasonably argue that at some level it doesn't matter -- that electronics components nor human ears are perfect and within that error we have a great deal of breathing room. But, no, you cannot create a digital signal that is an exact replica of an analog signal -- there aren't enough atoms in the universe to encode onto.
I have found my new true love. My wife likes them too so it's okay!
I have a big space to fill. Any reason why I should not go for two of these baby's and wire them up together? Is it as easy as splitting the audio signal, or would that be a no no to the audiophile?
Thanks!
Originally posted by hardeeharhar
But, no, you cannot create a digital signal that is an exact replica of an analog signal -- there aren't enough atoms in the universe to encode onto.
Whilst this is kind of true, it doesn't mean that vinyl has more information than CD.
CD does two things that vinyl doesn't: it samples the signal in time, and it quantises it in amplitude.
The time sampling limits the time resolution, and therefore gives an upper frequency limit. As I pointed out, the upper frequency limit is the same for CD as it is for vinyl, except any given vinyl record will decrease in bandwidth each time it is played.
If you take an analogue signal, and sample it at anything over twice the maximum frequency component, you can reconstruct that signal perfectly. This is a simple application of mathematics, it is a fact. Note, I have not said anything about quantising the signal.
Next, something a lot of people don't understand is that quantising is exactly the same as adding noise to the signal. But there is noise in a vinyl system, too. CD quantisation adds less noise to the signal than the vinyl recording and playback system. Noise is not information.
So, if you take an analogue signal, it will contain noise. There is absolutely no point in using more bits, just so you can accurately encode the noise, because you would need a perfectly noiseless playback system (impossible to achieve) in order to play back exactly what you recorded, and this would be pointless anyway, because the noise is, well, noise. It is not the signal.
Originally posted by Dutchy
Hey All,
I have a big space to fill. Any reason why I should not go for two of these baby's and wire them up together? Is it as easy as splitting the audio signal, or would that be a no no to the audiophile?
Thanks!
All the reports that I've read so far indicate that the iPod HiFi is well capable of filling a large space with sound. I don't think there's any way to wire two together (at least not if you want to source music from an iPod docked into the top of one of the HiFis).
The iPod HiFi does have an external input, however. So, you could make up a cable that had a stereo connector at one end, and four connectors at the other: right signal x 2 and left signal x 2. Connect the two right signals to one iPod HiFi, and the two left signals to the other iPod HiFi. Then all you have to do is synchronise the volume. The only problem is, you can only do that by ear, because the HiFi doesn't have a volume read-out. Also, I don't know if the HiFi works without an iPod docked into it.
It better work without the ipod. I am still a happy owner over ipod v1.0 with firewire connector! Only apple product I felt safe to early adopt. HiFi will be the second..
Originally posted by Dutchy
I have a big space (studio). I was looking to splurge on a set of these [ http://www.genelec.com/products/bi_amp.php ] some time this year.. Now I'm heading Apple's way again.
It better work without the ipod. I am still a happy owner over ipod v1.0 with firewire connector! Only apple product I felt safe to early adopt. HiFi will be the second..
These are probably very good speakers, but honestly they're
looking b**t-u*ly. What about pricing? (too lazy to google)
Originally posted by Vox Barbara
These are probably very good speakers, but honestly they're
looking b**t-u*ly. What about pricing? (too lazy to google)
HAHA! To each his own. They look ok in da flesh... or should I say cast aluminum. The weigh 15KG each. Sounded A M A Z I N G in the shop studio. Require a sub-woofer for big spaces. They have built in amplification. I think they were around 350 euro ($400) each. So, not cheap, but not crazy audiophile pricing. They are worth checking for sure. I'm going for a HiFi and will see how far that gets me.
Originally posted by Dutchy
HAHA! To each his own. They look ok in da flesh... or should I say cast aluminum. The weigh 15KG each. Sounded A M A Z I N G in the shop studio. Require a sub-woofer for big spaces. They have built in amplification. I think they were around 350 euro ($400) each. So, not cheap, but not crazy audiophile pricing. They are worth checking for sure. I'm going for a HiFi and will see how far that gets me.
Well, no offend here
Genelec is a brand you have to factor in, IF you DO WANT
and NEED extraordinary quality regarding sound. No question here.
But, (and this is strictly a personal point of view, and hence a matter
of my very own taste) i do prefer Tannoy in look and sound over
Genelec (and other comparable brands in this price range).
sic and cheers
to be pretty sound anyway. Settled? Cheers.
Originally posted by Mr. H
If you take an analogue signal, and sample it at anything over twice the maximum frequency component, you can reconstruct that signal perfectly.
Nyquist's law always seemed kind of like a guess to me in school.
If it is really true, then why does 24/96 digital audio sound so much better than 16/44.1? Why does SACD sound even better still?
I think that people that get caught up in the math of sound end up missing out on the good stuff. You just can't get a good sounding stereo if you rely on specs and math - CD sound is harsh, brittle and dull sounding, and so is solid state sound.
Originally posted by hardeeharhar
Splinemodel,
But, no, you cannot create a digital signal that is an exact replica of an analog signal -- there aren't enough atoms in the universe to encode onto.
You don't need to create a digital signal that is an exact replica. You just need the signals max frequency to be less than 1/2 the sampling frequency, and you can regenerate a perfect copy of the digitized analog source signal if you have an ideal-low-pass filter. Now, ideal-low-pass filters are hard to come by. That's why you oversample and use a good FIR filter within the digital domain.
I don't mean to pull rank, but do have a degree in electrical engineering, during which time I focused on signal processing, and after which time I started working in the RFID industry. I do know what I'm talking about.
Nyquist's law always seemed kind of like a guess to me in school.
Then you didn't pay attention. Nyquist's law can be proven mathematically (relatively easily). Note that it has been dubbed a "law," which suggests that it's not some kind of hack. Moving on, if our mathematics are wrong, then either side of an argument on mathematical principles is flawed, and you should give up.
I think that people that get caught up in the math of sound end up missing out on the good stuff. You just can't get a good sounding stereo if you rely on specs and math - CD sound is harsh, brittle and dull sounding, and so is solid state sound.
This is a dumb argument. If a digital signal can be reconstructed exactly into the source signal, then it will sound exactly the same as the source signal. The reason why it doesn't is because the system you're using doesn't do a good job of reconstructing. If it could, you wouldn't know the difference.
Incidentally, there's also plenty of scientific evidence suggesting the placebo effect is real. I think that's really the culprit here.
Originally posted by e1618978
Nyquist's law always seemed kind of like a guess to me in school.
No, it wasn't a guess. If you take any signal, and sample it at a rate at least double the highest frequency that signal contains, you have captured it perfectly. Take a signal x(t), and sample it in time, to give x[n]. There is one, and only one, signal that can be represented by x[n], and that's x(t). Therefore, it is possible to perfectly reconstruct x(t) from x[n]. (Yes, I know this isn't a proof, but it is the general idea and principle).
Originally posted by e1618978
If it is really true, then why does 24/96 digital audio sound so much better than 16/44.1? Why does SACD sound even better still?
See my previous explanation. Also, placebo is a distinct possibility.
Originally posted by e1618978
CD sound is harsh, brittle and dull sounding, and so is solid state sound.
listen, I'm not trying to tell you that CD or solid state electronics sound better. You may prefer vinyl and tubes, fair enough. But don't kid yourself that it's because such a system is giving you "more of the original sound"; it is not. It is giving you less, and what it is giving you is more distorted and compressed (in terms of dynamic range). It is these distortions and dynamic range compressions that give vinyl and tubes their warmer, less harsh sound. If you want to hear exactly what was recorded, unfortunately, that's impossible, but the closest thing you can get is a recording and playback system as I outlined a few posts ago.
Originally posted by Splinemodel
You don't need to create a digital signal that is an exact replica.
CD players, even cheap ones, can sound dramatically different from each other. Not only that, but CDs made now sound a lot better than CDs made in the early 80s.
Even with lossless compression, the iPod mini sounds really bad when you hook it up to a stereo, much worse than any CD player.
I don't think that your "exact replica" is as exact as you think it is.
Originally posted by 1984
I used to hear well beyond 20 -20,000 but it was a while since my last test.
I very much doubt that claim, as even newly born infants are hard pressed to reach those limits. What i'd gather you are actually 'hearing' are harmonics of the higher/lower frequencies. Second, third and perhaps fourth order harmonics, not the actually figure over or under the numbers stated.
Originally posted by ZoranS
I very much doubt that claim, as even newly born infants are hard pressed to reach those limits. What i'd gather you are actually 'hearing' are harmonics of the higher/lower frequencies. Second, third and perhaps fourth order harmonics, not the actually figure over or under the numbers stated.
It depends what he means by "well above". With hearing, just like anything else, there is a spread of capabilities within the population. A very, very small number of people really can hear beyond 20 kHz, just like a very very small number of people have an IQ of 170 or more, and a very very small number of people can run 100 m in less than 9 seconds.
Also, I don't know what you're talking about with the harmonics thing. I think it is possible that when he tested his hearing, and was driving a speaker with greater than 20 kHz, the system could have produced a tone below 20 kHz (i.e. a sub-harmonic) it really depends on the distortion mechanisms in the sytem he was using.
Originally posted by Splinemodel
You don't need to create a digital signal that is an exact replica. You just need the signals max frequency to be less than 1/2 the sampling frequency, and you can regenerate a perfect copy of the digitized analog source signal if you have an ideal-low-pass filter. Now, ideal-low-pass filters are hard to come by. That's why you oversample and use a good FIR filter within the digital domain.
I don't mean to pull rank, but do have a degree in electrical engineering, during which time I focused on signal processing, and after which time I started working in the RFID industry. I do know what I'm talking about.
Great! And I don't mean to pull rank on you, I use analog to digital transforms and every day as well and I am getting my third degree!
Now sit down.
I was simply pointing out that you are making approximations about what signals are relevant and those that aren't. You can claim the ability to obtain a good enough replica because you are limiting the frequency space you are interested in to the range of human hearing, which isn't exactly the same as all frequencies.
This fact is why Nyquist's Theorem isn't a law.
Originally posted by hardeeharhar
Great! And I don't mean to pull rank on you, I use analog to digital transforms and every day as well and I am getting my third degree!
Now sit down.
I was simply pointing out that you are making approximations about what signals are relevant and those that aren't. You can claim the ability to obtain a good enough replica because you are limiting the frequency space you are interested in to the range of human hearing, which isn't exactly the same as all frequencies.
Really? (about your qualifications), well this is strange post for you to make, then.
We are talking about vinyl vs. CD. They both have the same bandwidth (as long as you only play your record once). Therefore CD is losing nothing to vinyl in frequency terms.
Originally posted by hardeeharhar
This fact is why Nyquist's Theorem isn't a law.
?? The fact that you can't capture a signal with infinite frequency content by sampling it has got bugger all to do with whether Nyquist's Theorem is true or not. In fact, it is Nyquist's Theorem that states that you cannot capture a signal with infinite frequency content by sampling it.
Originally posted by Mr. H
We are talking about vinyl vs. CD. They both have the same bandwidth (as long as you only play your record once). Therefore CD is losing nothing to vinyl in frequency terms.
I am not talking about vinyl... I was trying to get splinemodel to be more precise in his language, that is all...
Originally posted by Mr. H
?? The fact that you can't capture a signal with infinite frequency content by sampling it has got bugger all to do with whether Nyquist's Theorem is true or not. In fact, it is Nyquist's Theorem that states that you cannot capture a signal with infinite frequency content by sampling it.
It isn't a question of whether it is true or not, it isn't a law because it doesn't cover all frequency space..
Originally posted by hardeeharhar
It isn't a question of whether it is true or not, it isn't a law because it doesn't cover all frequency space..
Well, I suppose that depends. Does infinite frequency space exist? And if it does, just sample it at infinity x 2 (yes, you can have different size infinities e.g. the number of irrational numbers >> the number of integers, but they are both infinite in number).