Hi Res Sound Quality - Page 4 - General HIFI Discussion - HIFICRITIC FORUM - HIFICRITIC FORUM : hi fi audio systems forum

Welcome Guest! To enable all features please Login or Register.

Notification

Error

HIFICRITIC FORUM » HIFI Discussion Forums » General HIFI Discussion » Hi Res Sound Quality

6 Pages«<2 345 6 >

Hi Res Sound Quality

Share Options

Previous Topic Next Topic

hifistan		#61 Posted : 19 January 2012 16:27:31(UTC) Retweet
Rank: Administrator Joined: 18/11/2008(UTC) Posts: 759 Location: Indiana, USA Was thanked: 5 time(s) in 3 post(s)		ashleym wrote: Perhaps you "shouldnt" hear the difference as you go up in sample frequency but the benefits will include moving the brickwall filter further away from the audio band. I believe this is less of an issue as the industry gets better at filter design. See the developments such as the Meridian and DCS Apodising filters. Listen to what happens when you put a supertweeter on a system. And read the article in the Critic (I dont have the copy to hand) on research around our ability to hear higher than traditionally expected frequencies. If that was the article on research done at the University of South Carolina the point of it was that timing, not frequency response, was critical. I.E.; that a 90 year old man retained the ability to judge the quality of sound even when the frequency response of his hearing was severely curtailed. And that we could hear timing differences of as little as 6 one millionth of a second. My own experiences as I age are along these same lines although unfortunately no double blind tests are practical until I repair the Tardis.


User Profile View All Posts by User View Thanks

Werner		#62 Posted : 23 January 2012 07:15:36(UTC) Retweet
Rank: HIFI Novice Joined: 28/04/2009(UTC) Posts: 77 Location: Belgium		bdiament wrote: Adding an additional conversion step means one is no longer comparing the different resolution files. I find ... an accurate assessment ... is possible only when a single variable is being tested. Yes. Which is exactly why one should apply off-line oversampling to the low-res file. The reason is that a given DAC chip behaves quite differently between 1x and 4x input rate. If you look into the architecture that 99% of the in-DAC digital filters out there use you'll see that they consist of a cascade of 2x filters that are engaged according to the input data rate: with 1x rate you get 3 filters in cascade (of which the first one is typically quite complex and steep, and the last one typically a half-assed job), while for 4x input rate you only get the last filter in action. Now the ideal replay filter is known, as it is imposed by the sampling theorem. No DAC chip implements this ideal. But software like iZotope comes pretty close. So if you use off-line SRC from 44.1kHz to, say, 192kHz, you give the low-res file something closer to ideal conversion, and during comparison with native 192kHz you force the DAC to operate in one and the same mode.
		WWW

User Profile View All Posts by User View Thanks

Sone		#63 Posted : 25 January 2012 15:53:40(UTC) Retweet
Rank: HIFI Newbie Joined: 20/12/2011(UTC) Posts: 2 Location: UK		Werner wrote: Which is exactly why one should apply off-line oversampling to the low-res file. The reason is that a given DAC chip behaves quite differently between 1x and 4x input rate. If you look into the architecture that 99% of the in-DAC digital filters out there use you'll see that they consist of a cascade of 2x filters that are engaged according to the input data rate: with 1x rate you get 3 filters in cascade (of which the first one is typically quite complex and steep, and the last one typically a half-assed job), while for 4x input rate you only get the last filter in action. Now the ideal replay filter is known, as it is imposed by the sampling theorem. No DAC chip implements this ideal. But software like iZotope comes pretty close. So if you use off-line SRC from 44.1kHz to, say, 192kHz, you give the low-res file something closer to ideal conversion, and during comparison with native 192kHz you force the DAC to operate in one and the same mode. Hi I'm about to try upsampling a few of my lowly MP3 files using the iZotope 64-bit SRC in Soundforge 10. I'm not sure whether to also convert to 24 bits (iZotope MBIT + Dither converter) or leave it at 16 bits? My burson DAC supports 24/ 192 Would doing all this give better results than upsampling on the fly with something like the soX resampler in Foobar? Thanks Edited by user 26 January 2012 14:38:30(UTC) \| Reason: Not specified


User Profile View All Posts by User View Thanks

ashleym		#64 Posted : 25 January 2012 21:58:33(UTC) Retweet
Rank: HIFI Veteran Joined: 02/02/2009(UTC) Posts: 901 Location: uk Thanks: 1 times Was thanked: 2 time(s) in 2 post(s)		Lowly MP3s will stay lowly MP3s. Yes they might sound a little bit better after SRC but my experience using Protools (typically Fs 16K up to 44.1K) for quite lo-res files it that there is only a little gain, we mainly do it to keep the files in a standard format. The files I deal with might be lower bandwidth than your MP3s so I cant guarantee what you will hear. If you want to hear SRC done badly get a copy of dBPoweramp.


User Profile View All Posts by User View Thanks

bdiament		#65 Posted : 27 January 2012 15:39:30(UTC) Retweet
Rank: HIFI Addict Joined: 09/03/2009(UTC) Posts: 196 Location: New York		Hi Werner, Werner wrote: Yes. Which is exactly why one should apply off-line oversampling to the low-res file. The reason is that a given DAC chip behaves quite differently between 1x and 4x input rate. If you look into the architecture that 99% of the in-DAC digital filters out there use you'll see that they consist of a cascade of 2x filters that are engaged according to the input data rate: with 1x rate you get 3 filters in cascade (of which the first one is typically quite complex and steep, and the last one typically a half-assed job), while for 4x input rate you only get the last filter in action. Now the ideal replay filter is known, as it is imposed by the sampling theorem. No DAC chip implements this ideal. But software like iZotope comes pretty close. So if you use off-line SRC from 44.1kHz to, say, 192kHz, you give the low-res file something closer to ideal conversion, and during comparison with native 192kHz you force the DAC to operate in one and the same mode. I have never heard 16/44 sound better (i.e., more like the master played in the application in which it was recorded) than via my ULN-8. Adding an additional conversion step will, in my view, take one further from, not closer to, an accurate comparison of the two formats, particularly since my preference is to listen to files at their native rate. Your test would make more sense to me if I preferred upsampling everything to 24/192 prior to listening. I hope we can agree to disagree on this. Best regards, Barry www.soundkeeperrecordings.com www.barrydiamentaudio.com
		WWW BLOG

User Profile View All Posts by User View Thanks

bencat		#66 Posted : 27 January 2012 16:23:59(UTC) Retweet
Rank: HIFI Addict Joined: 23/05/2009(UTC) Posts: 99 Location: Liverpool		Barry I respect your views on this and realise that you have considerably more real life experience of digital files including CD quality that I do. However I have found that for my own listening that I have come to prefer CD with the general application of oversampling applied. I understand your feeling and view that each medium we use should be played as close as possible to the original signal that was captured from the recirding process. Intellectually I have not other course but to agree as I do exactle the same when using my equipment , shortest signal path , shortest wire connections no tone controls. All of these are implemented to give as you say the best translation of the medium in it native form. Yet despite this I use a system that does change it and for me in my system what I do seems to work (that is not strictly true it also works for my wife who has no interest in my equipment only music and she says it currently sounds better). I have removed the upsampler unit out a number of times and for me the music deflates as if the air has been let out of it . Cymbals become a sizzle noise instead of a metalic ring . I can not explain this and I do not doubt your findings for yourself but as you say I think we are going to have to agree to disagree . One thing I think we both agree on is if you get your system working and it makes and compels you to lsten to more music that you must be doing something right .
		System Theta Data Basic II Transport , Perpetual Technologies PA-1 Upsampler, PA-3 Dac , Concordant Exhillirant Pre ,Krell KSA50 Power , Harbeth Compact II Monitors .


User Profile View All Posts by User View Thanks

bdiament		#67 Posted : 01 February 2012 16:36:49(UTC) Retweet
Rank: HIFI Addict Joined: 09/03/2009(UTC) Posts: 196 Location: New York		Hi bencat, bencat wrote: Barry I respect your views on this and realise that you have considerably more real life experience of digital files including CD quality that I do. However I have found that for my own listening that I have come to prefer CD with the general application of oversampling applied. I understand your feeling and view that each medium we use should be played as close as possible to the original signal that was captured from the recirding process. Intellectually I have not other course but to agree as I do exactle the same when using my equipment , shortest signal path , shortest wire connections no tone controls. All of these are implemented to give as you say the best translation of the medium in it native form. Yet despite this I use a system that does change it and for me in my system what I do seems to work (that is not strictly true it also works for my wife who has no interest in my equipment only music and she says it currently sounds better). I have removed the upsampler unit out a number of times and for me the music deflates as if the air has been let out of it . Cymbals become a sizzle noise instead of a metalic ring . I can not explain this and I do not doubt your findings for yourself but as you say I think we are going to have to agree to disagree . One thing I think we both agree on is if you get your system working and it makes and compels you to lsten to more music that you must be doing something right . We don't necessarily disagree. Well it depends first on whether or not we're talking about the same thing: I differentiate between oversampling (as used by the filter sections in the most transparent DACs I've heard) and upsampling (converting the sample rate of the file prior to feeding it to the DAC). Further, I differentiate between off-line upsampling (conversion before playing) and on-line upsampling (conversion during playback); to my ears, the latter is sonically damaging, even with the best conversion algorithms. As far as upsampling (off-line), depending on the algorithm used to upsample a CD and depending on the DAC and the filters it uses at different sample rates, there can in my experience, most definitely be advantage in applying the finest algorithms insomuch as the playback filtering is both gentler (less steep) and is moved further away from the top of the audible spectrum. Though I tend to listen to the files on my server at their native rates, one place I would most definitely avoid any upward sample rate alterations is when comparing high sample rates with Redbook (i.e., CD) sample rates. I prefer to compare only a single variable at a time; when I'm comparing 16/44 vs. 24/192, I want to listen to 16/44 vs. 24/192. To alter the former would mean I'd be comparing 16/44-24/192 to 24/192. I did my comparisons using the best 16/44 DAC and best 24/192 DAC in my experience (which luckily for me, occur in the same box). Another comparison we've done is to record the same event at both rates and to compare those. Again, when making such comparisons, I won't introduce an additional variable. That said, even when upsampling prior to listening, using the most transparent conversion algorithm I know of, 16/44-24/192 is, from my perspective, still not in the same universe as true 24/192. (It is a shame some on-line download services are selling 16/44-24/96 as "high res". Night Train, when in the finest crystal glass, will never - to me - taste like Dom Perignon. ;-}) In the end, I'm convinced we all hear things differently and have different sensitivities to different aspects of sound. While I must approach audio in the best way I see it, ultimately, I would never argue with what brings anyone their listening pleasure. Best regards, Barry www.soundkeeperrecordings.com www.barrydiamentaudio.com Edited by user 01 February 2012 16:39:44(UTC) \| Reason: Not specified
		WWW BLOG

User Profile View All Posts by User View Thanks

Werner		#68 Posted : 02 February 2012 07:54:23(UTC) Retweet
Rank: HIFI Novice Joined: 28/04/2009(UTC) Posts: 77 Location: Belgium		There seems to be a bit of confusion. bdiament wrote: Though I tend to listen to the files on my server at their native rates, Ignoring NOS systems (which are a minority anyway), no digital audio system or DAC renders playback at 'native rate'. Likewise, no ADC system truly converts from analogue at 'native rate'. bencat wrote: I have come to prefer CD with the general application of oversampling applied. I understand your view that each medium we use should be played as close as possible to the original signal that was captured from the recording process. ... I have removed the upsampler unit out a number of times and for me the music deflates as if the air has been let out of it . Oversampling, upsampling, in the DAC (or its preceding filter chip), on-line in a seperate box or in a computer, off-line on files, ... all are mathematically the very same thing. These things only differ in implementation details such as numerical accuracy. The ideal filter characteristic to which such a process should adhere has been well known for almost a century. Many strive to this ideal, but specifically on-chip filters are subject to the rules of economy and tend to cut some corners (please read ADC and DAC datasheets). In the software world some implementations are close to perfect. Others are abysmal, demonstrating the design team's incompetence when it comes to signal theory. Then, both in software and in hardware, there are a few that deliberately discard proper theory, claiming it wrong. Well, it isn't and they are, but in the end the only thing that matters is if you like it or not. (Then again, what also matters is that this subjective preference is not wrongly attributed to specific theories.) So, coming back to the quote above: if you remove the upsampling unit, then some other box in the system, probably the DAC, will invisibly take over its processing task, and you will not have removed that up/oversampling step. Only replaced one version with another. bdiament wrote: As far as upsampling (off-line), depending on the algorithm used to upsample a CD and depending on the DAC and the filters it uses at different sample rates, there can in my experience, most definitely be advantage in applying the finest algorithms insomuch as the playback filtering is both gentler (less steep) and is moved further away from the top of the audible spectrum. Coming from CD and fully working according to the book the compounded filtering has to be steep and has to be at 22kHz. And this is what nearly all implementations do. (If they didn't they would be tone controls.) The only difference is, again, which particular step of the filtering is done in which particular box or process, and to which specific accuracy. Edited by user 02 February 2012 07:57:17(UTC) \| Reason: Not specified
		WWW

User Profile View All Posts by User View Thanks

bdiament		#69 Posted : 03 February 2012 21:55:30(UTC) Retweet
Rank: HIFI Addict Joined: 09/03/2009(UTC) Posts: 196 Location: New York		Hi Werner, Werner wrote: There seems to be a bit of confusion. bdiament wrote: Though I tend to listen to the files on my server at their native rates, Ignoring NOS systems (which are a minority anyway), no digital audio system or DAC renders playback at 'native rate'. Likewise, no ADC system truly converts from analogue at 'native rate'. Sorry for any confusion. I am not talking about the oversampling filters in the converters. I'm talking about the sample rate of the file I feed to the converters. bdiament wrote: As far as upsampling (off-line), depending on the algorithm used to upsample a CD and depending on the DAC and the filters it uses at different sample rates, there can in my experience, most definitely be advantage in applying the finest algorithms insomuch as the playback filtering is both gentler (less steep) and is moved further away from the top of the audible spectrum. Werner wrote: Coming from CD and fully working according to the book the compounded filtering has to be steep and has to be at 22kHz. And this is what nearly all implementations do. (If they didn't they would be tone controls.) The only difference is, again, which particular step of the filtering is done in which particular box or process, and to which specific accuracy. Perhaps there is a communication lapse or I'm just not writing my thoughts clearly, for which I apologize. If I convert the sample rate of a 44.1k file to 192k, there will be no 22.05k filter in the playback; as far as the original 44.1k signal is concerned, the steep, brickwall filter for playback no longer exists. It has been replaced by one which will be much more benign at greater than 4x the frequency. With this, done off-line with the best SRC algorithms, my experience has been that there can be a clear audible advantage. However, this very much depends on the DAC used for playback. As I've mentioned elsewhere, I've found a number of DACs spec'd for 4x rates (176.4 and 192k) actually perform worse at these rates than they do at lower ones, which I attribute to their clocking and analog stage design. All, just my perspective of course. I understand and respect that you may experience it differently. Best regards, Barry www.soundkeeperrecordings.com www.barrydiamentaudio.com
		WWW BLOG

User Profile View All Posts by User View Thanks

kengale		#70 Posted : 03 February 2012 23:57:48(UTC) Retweet
Rank: HIFI Veteran Joined: 25/11/2008(UTC) Posts: 982 Location: UK Was thanked: 2 time(s) in 2 post(s)		bdiament wrote: Hi Werner, Werner wrote: There seems to be a bit of confusion. bdiament wrote: Though I tend to listen to the files on my server at their native rates, Ignoring NOS systems (which are a minority anyway), no digital audio system or DAC renders playback at 'native rate'. Likewise, no ADC system truly converts from analogue at 'native rate'. Sorry for any confusion. I am not talking about the oversampling filters in the converters. I'm talking about the sample rate of the file I feed to the converters. bdiament wrote: As far as upsampling (off-line), depending on the algorithm used to upsample a CD and depending on the DAC and the filters it uses at different sample rates, there can in my experience, most definitely be advantage in applying the finest algorithms insomuch as the playback filtering is both gentler (less steep) and is moved further away from the top of the audible spectrum. Werner wrote: Coming from CD and fully working according to the book the compounded filtering has to be steep and has to be at 22kHz. And this is what nearly all implementations do. (If they didn't they would be tone controls.) The only difference is, again, which particular step of the filtering is done in which particular box or process, and to which specific accuracy. Perhaps there is a communication lapse or I'm just not writing my thoughts clearly, for which I apologize. If I convert the sample rate of a 44.1k file to 192k, there will be no 22.05k filter in the playback; as far as the original 44.1k signal is concerned, the steep, brickwall filter for playback no longer exists. It has been replaced by one which will be much more benign at greater than 4x the frequency. With this, done off-line with the best SRC algorithms, my experience has been that there can be a clear audible advantage. However, this very much depends on the DAC used for playback. As I've mentioned elsewhere, I've found a number of DACs spec'd for 4x rates (176.4 and 192k) actually perform worse at these rates than they do at lower ones, which I attribute to their clocking and analog stage design. All, just my perspective of course. I understand and respect that you may experience it differently. Best regards, Barry www.soundkeeperrecordings.com www.barrydiamentaudio.com I think you've misunderstood the upsampling process - the upsampler alway has to include a nominal 22.05kHz filter that has exactly the same characteristics as the one which is normally inside an over-sampling DAC driven from the native 44.1kHz - otherwise you will get all the alias products present sampled at the higher frequency, but there all the same. All you've done is move the connection point between the upsampling filter and the final DAC function. These filters are usually designed to be around -120dB at 24.1kHz and around -60dB at the Nyquist frequency (22.05kHz), on the assumption that the filter on the original coding side will also be around -60dB at 22.05kHz, thus giving -120dB suppression of all aliasing products. Your statement "as far as the original 44.1k signal is concerned, the steep, brickwall filter for playback no longer exists" isn't true - it just exists in a different IC but in exactly the same place in the replay chain. And the best characteristic for this filter is a copy of the one used for coding - which of course will change from recording to recording according to which particular IC or algorithm was used on the recording side. As most recording these days is done with sample rates multiples of 48kHz, the filter characteristics are usually part of the sample rate converter used to change from nx48kHz to 44.1kHz. How you'd find out which one was used on a particular recording I've no idea.


User Profile View All Posts by User View Thanks

fas42		#71 Posted : 04 February 2012 04:52:01(UTC) Retweet
Rank: HIFI Novice Joined: 27/11/2011(UTC) Posts: 37 Location: NSW, Australia		kengale wrote: All you've done is move the connection point between the upsampling filter and the final DAC function. These filters are usually designed to be around -120dB at 24.1kHz and around -60dB at the Nyquist frequency (22.05kHz), on the assumption that the filter on the original coding side will also be around -60dB at 22.05kHz, thus giving -120dB suppression of all aliasing products. Your statement "as far as the original 44.1k signal is concerned, the steep, brickwall filter for playback no longer exists" isn't true - it just exists in a different IC but in exactly the same place in the replay chain. And the best characteristic for this filter is a copy of the one used for coding - which of course will change from recording to recording according to which particular IC or algorithm was used on the recording side. As most recording these days is done with sample rates multiples of 48kHz, the filter characteristics are usually part of the sample rate converter used to change from nx48kHz to 44.1kHz. How you'd find out which one was used on a particular recording I've no idea. There is a very large, very significant difference: the offline upsampling is causing any electrical interference from the IC filtering disturbing the key DAC digital to analogue conversion to be removed from the situation. And my own experiements have demonstrated to me that this is a considerable factor, far greater I believe, than cut off points and steepness of the filtering. Frank Edited by user 04 February 2012 04:53:09(UTC) \| Reason: Not specified


User Profile View All Posts by User View Thanks

kengale		#72 Posted : 04 February 2012 18:56:33(UTC) Retweet
Rank: HIFI Veteran Joined: 25/11/2008(UTC) Posts: 982 Location: UK Was thanked: 2 time(s) in 2 post(s)		fas42 wrote: kengale wrote: All you've done is move the connection point between the upsampling filter and the final DAC function. These filters are usually designed to be around -120dB at 24.1kHz and around -60dB at the Nyquist frequency (22.05kHz), on the assumption that the filter on the original coding side will also be around -60dB at 22.05kHz, thus giving -120dB suppression of all aliasing products. Your statement "as far as the original 44.1k signal is concerned, the steep, brickwall filter for playback no longer exists" isn't true - it just exists in a different IC but in exactly the same place in the replay chain. And the best characteristic for this filter is a copy of the one used for coding - which of course will change from recording to recording according to which particular IC or algorithm was used on the recording side. As most recording these days is done with sample rates multiples of 48kHz, the filter characteristics are usually part of the sample rate converter used to change from nx48kHz to 44.1kHz. How you'd find out which one was used on a particular recording I've no idea. There is a very large, very significant difference: the offline upsampling is causing any electrical interference from the IC filtering disturbing the key DAC digital to analogue conversion to be removed from the situation. And my own experiements have demonstrated to me that this is a considerable factor, far greater I believe, than cut off points and steepness of the filtering. Frank Fascinating - how did you seperate the oversampling section of an IC from the DAC part to tell that one was interfering with the other?


User Profile View All Posts by User View Thanks

Werner		#73 Posted : 06 February 2012 06:56:22(UTC) Retweet
Rank: HIFI Novice Joined: 28/04/2009(UTC) Posts: 77 Location: Belgium		kengale wrote: I think you've misunderstood the upsampling process - the upsampler alway has to include a nominal 22.05kHz filter that has exactly the same characteristics as the one which is normally inside an over-sampling DAC driven from the native 44.1kHz Exactly. kengale wrote: - otherwise you will get all the alias products 'images', actually. 'aliases' apply whenever the sampling rate is reduced. kengale wrote: These filters are usually designed to be around -120dB at 24.1kHz and around -60dB at the Nyquist frequency (22.05kHz), on the assumption that the filter on the original coding side will also be around -60dB at 22.05kHz, A bit of wishful thinking? For economincal reasons almost all hardware filters, and even many software types, are of the half-band type (or a cascade thereof), resulting in -6dB or -12dB at half the sample rate, not more. And stop band rejection in hardware filters seldomly exceeds 120dB, often is significantly worse. There is also no rule of assumption governing any symmetry between the recording and playback filter. The sampling theorem prescribes the perfect playback filter (and one can approximate it to arbitrary accuracy in software, but a HW version would be a tad complex), so that's nicely covered. On the other hand the theorem does not say anything meaningful about the recording filter. There may be room for trade-off there, but most recording side and downsampling filters in practice are, again, half-bands.
		WWW

User Profile View All Posts by User View Thanks

kengale		#74 Posted : 06 February 2012 13:29:34(UTC) Retweet
Rank: HIFI Veteran Joined: 25/11/2008(UTC) Posts: 982 Location: UK Was thanked: 2 time(s) in 2 post(s)		Werner wrote: kengale wrote: I think you've misunderstood the upsampling process - the upsampler alway has to include a nominal 22.05kHz filter that has exactly the same characteristics as the one which is normally inside an over-sampling DAC driven from the native 44.1kHz Exactly. kengale wrote: - otherwise you will get all the alias products 'images', actually. 'aliases' apply whenever the sampling rate is reduced. kengale wrote: These filters are usually designed to be around -120dB at 24.1kHz and around -60dB at the Nyquist frequency (22.05kHz), on the assumption that the filter on the original coding side will also be around -60dB at 22.05kHz, A bit of wishful thinking? For economincal reasons almost all hardware filters, and even many software types, are of the half-band type (or a cascade thereof), resulting in -6dB or -12dB at half the sample rate, not more. And stop band rejection in hardware filters seldomly exceeds 120dB, often is significantly worse. There is also no rule of assumption governing any symmetry between the recording and playback filter. The sampling theorem prescribes the perfect playback filter (and one can approximate it to arbitrary accuracy in software, but a HW version would be a tad complex), so that's nicely covered. On the other hand the theorem does not say anything meaningful about the recording filter. There may be room for trade-off there, but most recording side and downsampling filters in practice are, again, half-bands. Yep images not aliases! I was rather optimistic over the filter performance - extrapolating from the sort of filter performance I use on sonars but with much wider stop bands. For us image suppression is vital, as it can actually produce real ghost images on the final display, depending on the transmitted waveforms and subsequent correlation process. The playing filter matching the recording one is a thing I've seen in texts in quite a few places - The recording filter and playback filter are both FIR's rather than perfect filters, so that the replay impulse response is related to the cross-correlation product of the two filters. It's claimed that the best impulse response is the autocorrelation function of the recording one, though whether you could actually hear this I have no idea!


User Profile View All Posts by User View Thanks

Old Brock		#75 Posted : 09 February 2012 17:36:56(UTC) Retweet
Rank: HIFI Addict Joined: 19/10/2008(UTC) Posts: 131 Location: Hampshire		Sorry, I can't accept this. Lower bits is exactly equivalent to lower signal/noise ratio in analogue recording, 36dB down means on a decent tape deck that you only have about 35dB dynamic range of tape to play with, but no-one's complaining about this! I've listened to CD classical test pieces at 0db, then attenuated -60dB on a separate track, and all the key elements of the sound are still there, with good ambience recovery. If you tried this trick with tape you would be deafened by the tape hiss!! That notorious thin CD sound is all about faulty playback, the Achilles Heel of the medium, a type of disortion typically hard to eradicate; but once you experience the absence of it you wonder what all the fuss about vinyl vs. digital is about ... Frank[/quote] What on earth are you talking about?


User Profile View All Posts by User View Thanks

fas42		#76 Posted : 10 February 2012 01:56:00(UTC) Retweet
Rank: HIFI Novice Joined: 27/11/2011(UTC) Posts: 37 Location: NSW, Australia		Old Brock wrote: What on earth are you talking about? What part of what I said doesn't make sense? Frank


User Profile View All Posts by User View Thanks

bdiament		#77 Posted : 15 February 2012 16:03:03(UTC) Retweet
Rank: HIFI Addict Joined: 09/03/2009(UTC) Posts: 196 Location: New York		Hi kengale, kengale wrote: I think you've misunderstood the upsampling process - the upsampler alway has to include a nominal 22.05kHz filter... I stand corrected. Thank you for providing the opportunity to learn something new. A conversation with the designer of my preferred SRC algorithm (iZotope's "64-bit SRC") confirmed this. This particular algorithm allows adjustment of the filtering too, so one can hear the effects on the outcome. Thanks again (to Werner too). Best regards, Barry www.soundkeeperrecordings.com www.barrydiamentaudio.com
		WWW BLOG

User Profile View All Posts by User View Thanks

Shadders		#78 Posted : 15 February 2012 18:03:02(UTC) Retweet
Rank: HIFI Addict Joined: 07/04/2010(UTC) Posts: 170 Location: UK		Hmmm, I looked at the wiki page on upsampling - and effectively, the interpolation method provides the upsampled sequence of information at the new higher sample rate without the need for a filter. I think this means that for the 44.1kHz signal, you do not need to pass through a digital filter first, then upsample, since the interpolation provides this function already. Sample rate conversion - which can be up or down, i would expect to mean that the final sample frequency is not guaranteed to be an integer multiple or fraction of the initial sample rate. Such as an upsample 44.1kHz to 192kHz, or 176.4kHz downsample to 48kHz. Oversampling - i would expect to mean that the final sample rate is an integer multiple of the original signal sample rate - such as 44.1kHz oversampled to 176.4kHz. If interpolation is used, then no need for a digital filter applied to the incoming data before upsampling ? Regards, Richard. Edited by user 15 February 2012 18:03:37(UTC) \| Reason: Not specified


User Profile View All Posts by User View Thanks

bdiament		#79 Posted : 15 February 2012 20:36:38(UTC) Retweet
Rank: HIFI Addict Joined: 09/03/2009(UTC) Posts: 196 Location: New York		Hi Richard, Shadders wrote: Hmmm, I looked at the wiki page on upsampling - and effectively, the interpolation method provides the upsampled sequence of information at the new higher sample rate without the need for a filter. I think this means that for the 44.1kHz signal, you do not need to pass through a digital filter first, then upsample, since the interpolation provides this function already. Sample rate conversion - which can be up or down, i would expect to mean that the final sample frequency is not guaranteed to be an integer multiple or fraction of the initial sample rate. Such as an upsample 44.1kHz to 192kHz, or 176.4kHz downsample to 48kHz. Oversampling - i would expect to mean that the final sample rate is an integer multiple of the original signal sample rate - such as 44.1kHz oversampled to 176.4kHz. If interpolation is used, then no need for a digital filter applied to the incoming data before upsampling ? Regards, Richard. This is what I thought at first but I got the word from the person who designed the sample rate conversion algorithm I use (easily the most transparent of the dozens I've used - with the sole criterion for transparence being the unconverted original). He tells me there will indeed be a 22.05 kHz filter when converting a 44.1k original to 192k. Best regards, Barry www.soundkeeperrecordings.com www.barrydiamentaudio.com
		WWW BLOG

User Profile View All Posts by User View Thanks

Shadders		#80 Posted : 15 February 2012 21:13:09(UTC) Retweet
Rank: HIFI Addict Joined: 07/04/2010(UTC) Posts: 170 Location: UK		bdiament wrote: Hi Richard, Shadders wrote: Hmmm, I looked at the wiki page on upsampling - and effectively, the interpolation method provides the upsampled sequence of information at the new higher sample rate without the need for a filter. I think this means that for the 44.1kHz signal, you do not need to pass through a digital filter first, then upsample, since the interpolation provides this function already. Sample rate conversion - which can be up or down, i would expect to mean that the final sample frequency is not guaranteed to be an integer multiple or fraction of the initial sample rate. Such as an upsample 44.1kHz to 192kHz, or 176.4kHz downsample to 48kHz. Oversampling - i would expect to mean that the final sample rate is an integer multiple of the original signal sample rate - such as 44.1kHz oversampled to 176.4kHz. If interpolation is used, then no need for a digital filter applied to the incoming data before upsampling ? Regards, Richard. This is what I thought at first but I got the word from the person who designed the sample rate conversion algorithm I use (easily the most transparent of the dozens I've used - with the sole criterion for transparence being the unconverted original). He tells me there will indeed be a 22.05 kHz filter when converting a 44.1k original to 192k. Best regards, Barry www.soundkeeperrecordings.com www.barrydiamentaudio.com Hi Barry, Hmmm, i have a text by Analog Devices and a DSP book which seem to indicate that no 22.05kHz filter is required - so is it that the data that the person states requires a 22.05kHz is raw 44.1kHz sampled data with no analogue filter at the front end before sampling ? If you interpolate correctly between the 44.1kHz samples at a higher rate - then this will not introduce aliasing etc. (i think) Regards, Richard.


User Profile View All Posts by User View Thanks

Users browsing this topic
Guest (2)

6 Pages«<2 345 6 >

HIFICRITIC FORUM » HIFI Discussion Forums » General HIFI Discussion » Hi Res Sound Quality

Forum Jump

You cannot post new topics in this forum.
You cannot reply to topics in this forum.
You cannot delete your posts in this forum.
You cannot edit your posts in this forum.
You cannot create polls in this forum.
You cannot vote in polls in this forum.