|
|
Post by PaulBe on Jan 29, 2020 10:43:37 GMT -5
I would love to see actual blind listening test data on a recording from the same master in 24bit 48khz vs 24bit 192khz and if people can actually hear a difference. I don't believe there is such a study. Here is a recent article by Mark Waldrep is a proponent and skeptic of HD Audio. Personally, I think its all just about the reconstruction filter, affects on peaks (as @keithl stated) which like likely due to alterations in phases. It sure as hell isn't some uncanny ability to extremely low volume, highly directional content about 20kHz using our bone structure or other such non-sense. I'll believe that right after I start seeing UV. - Rich Here is an unblind study: Fixing a Hole Where the Rains Gets In by Mark Waldrep www.realhd-audio.com/?p=6774I don't believe Mark Waldrep is a sceptic of HD Audio, at least where the source material resolution and production quality can justify it. He creates his own HD Audio recordings. I have a few. They are fine recordings and music. He does suggest reasonable limits. I suggest Mark's fine Acoustica recordings: Goldberg Variations and Pachelbel Canon. Recorded in Dolby TrueHD, Atmos, 24 bit, 96K. Reproducing recordings like these, manipulated with 48K Dirac, would make no sense to me. |
|
|
|
Post by mikoz on Jan 29, 2020 11:11:10 GMT -5
I would love to see actual blind listening test data on a recording from the same master in 24bit 48khz vs 24bit 192khz and if people can actually hear a difference. I don't believe there is such a study. Here is a recent article by Mark Waldrep is a proponent and skeptic of HD Audio. Personally, I think its all just about the reconstruction filter, affects on peaks (as @keithl stated) which like likely due to alterations in phases. It sure as hell isn't some uncanny ability to extremely low volume, highly directional content about 20kHz using our bone structure or other such non-sense. I'll believe that right after I start seeing UV. - Rich
I don't understand the logic of many here... why ask for a study of what others hear on their equipment when you can just do it yourself and make up your own mind? You just need to find a good track (e.g. hdtracks.com) at 96/24 that's a music genre you like, make a copy of it and downconvert to 48/24, then have someone play it over and over again for you in the environment you normally listen to.
For me, I originally did this with my headphone setup using my LTA tube amp, two different DACs (Chord mojo and Sonica, one setup for unbalanced, one balanced) and an HD800S with Kardas clear. The track was a vinyl 96/24 mastering and I could pick out the difference 20/20 times. I converted the 96/24 FLAC to 48/24 on a PC.
|
|
KeithL
Administrator 
Posts: 10,517
|
Post by KeithL on Jan 29, 2020 11:17:36 GMT -5
To me, the biggest issue is trying to differentiate between "one is better" and "they're just a tiny bit different". Virtually all conversions involve some filtering... and filter differences can be extremely subtle.
Therefore, pretty much, ANY time you perform a conversion, tiny differences will be introduced.
I recall an example from several years ago (it was posted on the forums on the website for a "mastering and mixing magazine" - I think it was S.O.S.) Someone took a short clip from an unedited live recording that was originally recorded at DSD128. They then converted that clip to 24/96k PCM using what are widely agreed to be the two best conversion programs available (Korg Audiogate and Weiss Saracon). The purpose was to compare the sound signature of the two conversion programs.
Both conversion programs have multiple options - and both were run at their defaults. They posted both 24/96k PCM output samples for comparison. Both sounded excellent and, at first listen, sounded "pretty much the same". However, on careful listening, there were tiny differences. We're talking differences of the degree: "At about +14 seconds a small triangle is struck three times... on one sample the second and third strikes are equally loud but, on the other, the second strike is slightly louder than the third"
These are the sort of differences that you probably wouldn't notice if they weren't pointed out to you... and there's no way to know which is "right". (They are also the sort of tiny differences that you might reasonably believe could be due to nothing more than the power of suggestion.)
However, in this case, they are real differences, and you can see them if you zoom in on that part of the waveform in your editor. (Of course, they could just be due to slight differences in the default settings of the two conversion programs, so you might be able to choose different settings where they performed identically.)
To be quite honest.... if worrying about such subtle differences seems silly to you.... then you probably shouldn't worry about them.
However, if you want to experiment with various filters, and various conversion options, and various oversampling options... Then you should check out a program called "HQ Player" - sold by a company called Signalyst. It isn't cheap.. but it actually allows you to choose different filter settings, different oversampling algorithms, and such things, IN SOFTWARE, and compare them for yourself. Also note that some of the options require a very powerful computer - if you want to avoid stuttering and dropouts.
(I played with it several years ago and, yes, there are tiny but audible differences.... )
I would love to see actual blind listening test data on a recording from the same master in 24bit 48khz vs 24bit 192khz and if people can actually hear a difference. Why don’t you just try it and make up your own mind? Just get a 96/24 track and then copy it to a 48/24 track and get someone to play both over and over again? For some you’re correct, nothing will be audibly different. For others they’ll be able to quickly identify the difference. It will come down to the equipment you’re using and your ears. Most people who collect hires files have done this sort of thing at least in some capacity and have already made up there minds based on their own findings. |
|
|
|
Post by cwt on Jan 29, 2020 11:25:51 GMT -5
Apparently it does not. If the consumer can’t use 96khz then it doesn’t have enough power, it’s that simple. There’s really no comfort in knowing that it was originally planned to operate like that and now cannot... how does that help make you feel better that the processor is powerful enough? another question we are potentially skipping, how, many channels will Dirac operate on? Will we even get this on all channels at once (I feel like you have to be specific and ask the silly questions and hopefully, eventually,get a direct answer)? Didn't say it was supposed to make anyone feel better ;was just emphasising the griffin lites were not the limiting factor here in reply to "assuming that the current system does not have the power to get to that 96khz level)?"in other words perspective  The griffin lites have about 4 times the processing power of the previous falcon chips that restricted chipset avrs and pre pro's to 7.1.4 I once read and they are dual core  Dan once said that Dirac would be covering not just the bed channels but ceiling too . I think that should be de rigeur as dirac operates in both the space and time domains with late reverberations unlike audyssey who also got it wrong with 48khz processing apparently  |
|
|
|
Post by mikoz on Jan 29, 2020 11:33:56 GMT -5
To me, the biggest issue is trying to differentiate between "one is better" and "they're just a tiny bit different". Virtually all conversions involve some filtering... and filter differences can be extremely subtle.
Therefore, pretty much, ANY time you perform a conversion, tiny differences will be introduced.
I recall an example from several years ago (it was posted on the forums on the website for a "mastering and mixing magazine" - I think it was S.O.S.) Someone took a short clip from an unedited live recording that was originally recorded at DSD128. They then converted that clip to 24/96k PCM using what are widely agreed to be the two best conversion programs available (Korg Audiogate and Weiss Saracon). The purpose was to compare the sound signature of the two conversion programs.
Both conversion programs have multiple options - and both were run at their defaults. They posted both 24/96k PCM output samples for comparison. Both sounded excellent and, at first listen, sounded "pretty much the same". However, on careful listening, there were tiny differences. We're talking differences of the degree: "At about +14 seconds a small triangle is struck three times... on one sample the second and third strikes are equally loud but, on the other, the second strike is slightly louder than the third"
These are the sort of differences that you probably wouldn't notice if they weren't pointed out to you... and there's no way to know which is "right". (They are also the sort of tiny differences that you might reasonably believe could be due to nothing more than the power of suggestion.)
However, in this case, they are real differences, and you can see them if you zoom in on that part of the waveform in your editor. (Of course, they could just be due to slight differences in the default settings of the two conversion programs, so you might be able to choose different settings where they performed identically.)
To be quite honest.... if worrying about such subtle differences seems silly to you.... then you probably shouldn't worry about them.
However, if you want to experiment with various filters, and various conversion options, and various oversampling options... Then you should check out a program called "HQ Player" - sold by a company called Signalyst. It isn't cheap.. but it actually allows you to choose different filter settings, different oversampling algorithms, and such things, IN SOFTWARE, and compare them for yourself. Also note that some of the options require a very powerful computer - if you want to avoid stuttering and dropouts.
(I played with it several years ago and, yes, there are tiny but audible differences.... )
Why don’t you just try it and make up your own mind? Just get a 96/24 track and then copy it to a 48/24 track and get someone to play both over and over again? For some you’re correct, nothing will be audibly different. For others they’ll be able to quickly identify the difference. It will come down to the equipment you’re using and your ears. Most people who collect hires files have done this sort of thing at least in some capacity and have already made up there minds based on their own findings.
Keith, how about we stick to what the RMC-1 first ... seems like some basic questions are still on the table: 1. what is the sampling rate? 2. how many channels will be running at once at that rate. Why do I have a feeling that even 16 channels @ 48khz may be in question...
3. will you "enable" users to buy an upgrade module only for the RMC-1 to get back to what was originally promised for the sampling rate?
|
|
|
|
Post by fbczar on Jan 29, 2020 11:36:51 GMT -5
If the Dirac sample rate on the RMC-1 and XMC-2 is going to be no better than 48kHz it can only be viewed as a significant downgrade in performance for those of us with high res music collections, relative to the performance that was clearly touted. Say what you want, but 96KHz was the minimum expected sample rate for both processors and failure to meet that specification is unfair to customers. I traded in an XMC-1 for an XMC-2 and I would not have done so if I had known the Dirac sample rate would not have been upgraded. Unfortunately, those in the trade in program cannot recover if Emotiva does not live up to an upgraded Dirac sample rate. I think its fair to be upset about the sample rate, but are you honestly saying that you would have kept the XMC-1 and not upgraded to the XMC-2 with Atmos, 16 channels, new DAC's etc for the small cost it was (depending on the HDMI board you had) just because Dirac will be at 48khz? Yep. It is all about priorities. Everyone has their own. I would not have spent $1,050.00 just for Atmos. The new DACs are certainly not worth the expense. Implementing Atmos requires the considerable expense of new amplification and speakers. DSD over USB and Dirac with a higher sampling rate were my priorities. I wanted to upgrade. I was not interested in a lateral move. At this point I am also displeased that I will not get what I paid for. So much for trust. |
|
|
|
Post by mikoz on Jan 29, 2020 11:38:38 GMT -5
Apparently it does not. If the consumer can’t use 96khz then it doesn’t have enough power, it’s that simple. There’s really no comfort in knowing that it was originally planned to operate like that and now cannot... how does that help make you feel better that the processor is powerful enough? another question we are potentially skipping, how, many channels will Dirac operate on? Will we even get this on all channels at once (I feel like you have to be specific and ask the silly questions and hopefully, eventually,get a direct answer)? Didn't say it was supposed to make anyone feel better ;was just emphasising the griffin lites were not the limiting factor here in reply to "assuming that the current system does not have the power to get to that 96khz level)?"in other words perspective The griffin lites have about 4 times the processing power of the previous falcon chips that restricted chipset avrs and pre pro's to 7.1.4 I once read and they are dual core Dan once said that Dirac would be covering not just the bed channels but ceiling too . I think that should be de rigeur as dirac operates in both the space and time domains unlike audyssey who also got it wrong with 48khz processing apparently
I am pretty sure there's more to a prepro than just the DSP, there's a CPU in there too, onboard memory, etc. It's basically an embedded linux setup with a DSP chip, as such as it has a central CPU (probably something like an ARM or Xeon), probably some type of DDR memory, etc. Limitations can be found not just in the DSP itself...
|
|
|
|
Post by thxultra on Jan 29, 2020 12:22:31 GMT -5
I don't believe there is such a study. Here is a recent article by Mark Waldrep is a proponent and skeptic of HD Audio. Personally, I think its all just about the reconstruction filter, affects on peaks (as @keithl stated) which like likely due to alterations in phases. It sure as hell isn't some uncanny ability to extremely low volume, highly directional content about 20kHz using our bone structure or other such non-sense. I'll believe that right after I start seeing UV. - Rich I don't understand the logic of many here... why ask for a study of what others hear on their equipment when you can just do it yourself and make up your own mind? You just need to find a good track (e.g. hdtracks.com) at 96/24 that's a music genre you like, make a copy of it and downconvert to 48/24, then have someone play it over and over again for you in the environment you normally listen to.
For me, I originally did this with my headphone setup using my LTA tube amp, two different DACs (Chord mojo and Sonica, one setup for unbalanced, one balanced) and an HD800S with Kardas clear. The track was a vinyl 96/24 mastering and I could pick out the difference 20/20 times. I converted the 96/24 FLAC to 48/24 on a PC.
I actually do plan on doing this with my own equipment. I'm in the process of building my streaming/ Hi-res player setup. I had a Marantz before that had a builtin hi-res player but the XMC-2 doesn't. I recently got Roon and I am going to setup a Raspberry Pi with Ropiee to stream and play back hi-res files. I think however to get a real answer you have to run multiple tests with files created from the source and not converted. Files created using the same hardware also. I don't know for sure how Hd-tracks creates their files do they go back to the source for each different resolution or do they just convert the file? Does the conversion "change" the sound in anyway? Lots of variables this is why I am for a actual scientific study. Create new files from the source with the same hardware. |
|
|
|
Post by SOWK on Jan 29, 2020 12:30:54 GMT -5
Bottom line... The sample rate will be 48Khz in Dirac...
So stop asking that question you know they will refuse to answer!
LOL
|
|
|
|
Post by mikoz on Jan 29, 2020 12:31:27 GMT -5
I don't understand the logic of many here... why ask for a study of what others hear on their equipment when you can just do it yourself and make up your own mind? You just need to find a good track (e.g. hdtracks.com) at 96/24 that's a music genre you like, make a copy of it and downconvert to 48/24, then have someone play it over and over again for you in the environment you normally listen to.
For me, I originally did this with my headphone setup using my LTA tube amp, two different DACs (Chord mojo and Sonica, one setup for unbalanced, one balanced) and an HD800S with Kardas clear. The track was a vinyl 96/24 mastering and I could pick out the difference 20/20 times. I converted the 96/24 FLAC to 48/24 on a PC.
I actually do plan on doing this with my own equipment. I'm in the process of building my streaming/ Hi-res player setup. I had a Marantz before that had a builtin hi-res player but the XMC-2 doesn't. I recently got Roon and I am going to setup a Raspberry Pi with Ropiee to stream and play back hi-res files. I think however to get a real answer you have to run multiple tests with files created from the source and not converted. Files created using the same hardware also. I don't know for sure how Hd-tracks creates their files do they go back to the source for each different resolution or do they just convert the file? Does the conversion "change" the sound in anyway? Lots of variables this is why I am for a actual scientific study. Create new files from the source with the same hardware.
Well, just keep in mind that you won't have any say in the matter on the RMC-1 when you pump in a 96/24 file and it gets decimated. You will get what you get, so all you can do upfront is make up your own mind if your equipment and ears are going to miss the difference... if they are, then you should just assume that you'll miss it with Dirac in the absence of the ability to actually try it.
There was no difference in volume or any other artifacts I can detect in the conversion. I think some tracks you can buy from hdtracks, for example, at different sampling rates if you're worried about your conversion being the issue. Not sure if they will offer both 96/24 and 48/24 (they probably don't even go that low) but it's common to find 192/24 and 96/24 being offered.
|
|
KeithL
Administrator
Posts: 10,517
|
Post by KeithL on Jan 29, 2020 12:34:59 GMT -5
I would suggest that the justification is "future proofing".
You would hate to purchase a bunch of music recorded at 16/44k because you hear no difference on your current equipment... Only to find out that, after some future equipment upgrade, the difference is now obvious...
Historically... Many people have amassed large collections in MP3 or AAC format (iTunes)... Only to start noticing the flaws in them after purchasing new gear... And ending up purchasing new copies of stuff they'd already bought.
So you do want to be at least reasonably sure that the format you standardize on will serve you both now and in the future. I don't understand the logic of many here... why ask for a study of what others hear on their equipment when you can just do it yourself and make up your own mind? You just need to find a good track (e.g. hdtracks.com) at 96/24 that's a music genre you like, make a copy of it and downconvert to 48/24, then have someone play it over and over again for you in the environment you normally listen to.
For me, I originally did this with my headphone setup using my LTA tube amp, two different DACs (Chord mojo and Sonica, one setup for unbalanced, one balanced) and an HD800S with Kardas clear. The track was a vinyl 96/24 mastering and I could pick out the difference 20/20 times. I converted the 96/24 FLAC to 48/24 on a PC.
I actually do plan on doing this with my own equipment. I'm in the process of building my streaming/ Hi-res player setup. I had a Marantz before that had a builtin hi-res player but the XMC-2 doesn't. I recently got Roon and I am going to setup a Raspberry Pi with Ropiee to stream and play back hi-res files. I think however to get a real answer you have to run multiple tests with files created from the source and not converted. Files created using the same hardware also. I don't know for sure how Hd-tracks creates their files do they go back to the source for each different resolution or do they just convert the file? Does the conversion "change" the sound in anyway? Lots of variables this is why I am for a actual scientific study. Create new files from the source with the same hardware. |
|
|
|
Post by TDifEQ on Jan 29, 2020 12:38:52 GMT -5
With DL and 4k dv/atmos disc movies, I expect better envelopment, rumble/lfe effects, better sounding music highs, clearer mid range and tighter sound effects, especially in John Wick and Transformers movies, with 9.1.6 config. plus BEQ. Today, my 9.1.6 config sounds very good. Can't wait.
|
|
|
|
Post by thxultra on Jan 29, 2020 12:41:01 GMT -5
I would suggest that the justification is "future proofing". You would hate to purchase a bunch of music recorded at 16/44k because you hear no difference on your current equipment... Only to find out that, after some future equipment upgrade, the difference is now obvious... Historically... Many people have amassed large collections in MP3 or AAC format (iTunes)... Only to start noticing the flaws in them after purchasing new gear... And ending up purchasing new copies of stuff they'd already bought.
So you do want to be at least reasonably sure that the format you standardize on will serve you both now and in the future. I actually do plan on doing this with my own equipment. I'm in the process of building my streaming/ Hi-res player setup. I had a Marantz before that had a builtin hi-res player but the XMC-2 doesn't. I recently got Roon and I am going to setup a Raspberry Pi with Ropiee to stream and play back hi-res files. I think however to get a real answer you have to run multiple tests with files created from the source and not converted. Files created using the same hardware also. I don't know for sure how Hd-tracks creates their files do they go back to the source for each different resolution or do they just convert the file? Does the conversion "change" the sound in anyway? Lots of variables this is why I am for a actual scientific study. Create new files from the source with the same hardware. This is something I have found I can really hear a huge difference between MP3 and FLAC files on my XMC-2. The difference is much greater then it was on my Marantz. |
|
|
|
Post by thxultra on Jan 29, 2020 12:43:00 GMT -5
I actually do plan on doing this with my own equipment. I'm in the process of building my streaming/ Hi-res player setup. I had a Marantz before that had a builtin hi-res player but the XMC-2 doesn't. I recently got Roon and I am going to setup a Raspberry Pi with Ropiee to stream and play back hi-res files. I think however to get a real answer you have to run multiple tests with files created from the source and not converted. Files created using the same hardware also. I don't know for sure how Hd-tracks creates their files do they go back to the source for each different resolution or do they just convert the file? Does the conversion "change" the sound in anyway? Lots of variables this is why I am for a actual scientific study. Create new files from the source with the same hardware. Well, just keep in mind that you won't have any say in the matter on the RMC-1 when you pump in a 96/24 file and it gets decimated. You will get what you get, so all you can do upfront is make up your own mind if your equipment and ears are going to miss the difference... if they are, then you should just assume that you'll miss it with Dirac in the absence of the ability to actually try it.
There was no difference in volume or any other artifacts I can detect in the conversion. I think some tracks you can buy from hdtracks, for example, at different sampling rates if you're worried about your conversion being the issue. Not sure if they will offer both 96/24 and 48/24 (they probably don't even go that low) but it's common to find 192/24 and 96/24 being offered.
This will only happen if you have Direc engaged however. If you listen in pure direct mode this will not be the case you will still get 96/24. You are actually agreeing with Keith here and saying listen to your ears. If it sounds better with Direc on then listen to it that way... |
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Jan 29, 2020 13:52:09 GMT -5
Soon: Definition- Soon = Occuring within a brief period of time. eg. "Mom - when's dinner?" answer "Soon."
...Emo's definition = Something that might occur in the near future, but only after a great deal of money and effort is thrown at it  . Which then leads to the necessity of clarifying what is meant by "near future", and who's "future" and what's meant by "near". Maybe we all misunderstood and they meant "in the 'year' future." 
Being one of the first RMC-1 owners has had both high and low points. Sounds great but... , broken unit - fixed with new, firmware that fixes and breaks stuff. When the unit works it's fantastic and when it acts up 1 year + after I first started I just want to give up... then another firmware update and things are looking up (1.7.6 is very good). Last week, I finally put the Dirac calibration mic and misc. stuff from my first unboxing in storage... waiting for "the Near future." |
|
KeithL
Administrator
Posts: 10,517
|
Post by KeithL on Jan 29, 2020 14:01:31 GMT -5
I'm going to give you some PROVISIONAL answers here..... (that means I'm guessing)...
Having Dirac run on all channels is a high priority so I expect that will be the case. I expect Dirac to initially run at 48k (but I cannot rule out the possibility that might change later). It is unlikely that any sort of hardware upgrade will be offered later specifically to increase the sample rate at which Dirac operates.
To me, the biggest issue is trying to differentiate between "one is better" and "they're just a tiny bit different". Virtually all conversions involve some filtering... and filter differences can be extremely subtle.
Therefore, pretty much, ANY time you perform a conversion, tiny differences will be introduced.
I recall an example from several years ago (it was posted on the forums on the website for a "mastering and mixing magazine" - I think it was S.O.S.) Someone took a short clip from an unedited live recording that was originally recorded at DSD128. They then converted that clip to 24/96k PCM using what are widely agreed to be the two best conversion programs available (Korg Audiogate and Weiss Saracon). The purpose was to compare the sound signature of the two conversion programs.
Both conversion programs have multiple options - and both were run at their defaults. They posted both 24/96k PCM output samples for comparison. Both sounded excellent and, at first listen, sounded "pretty much the same". However, on careful listening, there were tiny differences. We're talking differences of the degree: "At about +14 seconds a small triangle is struck three times... on one sample the second and third strikes are equally loud but, on the other, the second strike is slightly louder than the third"
These are the sort of differences that you probably wouldn't notice if they weren't pointed out to you... and there's no way to know which is "right". (They are also the sort of tiny differences that you might reasonably believe could be due to nothing more than the power of suggestion.)
However, in this case, they are real differences, and you can see them if you zoom in on that part of the waveform in your editor. (Of course, they could just be due to slight differences in the default settings of the two conversion programs, so you might be able to choose different settings where they performed identically.) To be quite honest.... if worrying about such subtle differences seems silly to you.... then you probably shouldn't worry about them.
However, if you want to experiment with various filters, and various conversion options, and various oversampling options... Then you should check out a program called "HQ Player" - sold by a company called Signalyst. It isn't cheap.. but it actually allows you to choose different filter settings, different oversampling algorithms, and such things, IN SOFTWARE, and compare them for yourself. Also note that some of the options require a very powerful computer - if you want to avoid stuttering and dropouts.
(I played with it several years ago and, yes, there are tiny but audible differences.... )
Keith, how about we stick to what the RMC-1 first ... seems like some basic questions are still on the table: 1. what is the sampling rate? 2. how many channels will be running at once at that rate. Why do I have a feeling that even 16 channels @ 48khz may be in question...
3. will you "enable" users to buy an upgrade module only for the RMC-1 to get back to what was originally promised for the sampling rate?
|
|
klinemj
Emo VIPs 
Official Emofest Scribe
Posts: 15,406
|
Post by klinemj on Jan 29, 2020 14:30:28 GMT -5
I expect Dirac to initially run at 48k (but I cannot rule out the possibility that might change later). Question for you Keith...at the last Emofest (not the recent tech day, but at the last full Emofest), I recall Dan and Lonnie saying they built the RMC-1 to be powerful enough to run 96k, but what was ultimately be delivered was up to Dirac. Is that what's in play here as it relates to your comment? Thanks, Mark |
|
|
|
Post by andersmi on Jan 29, 2020 14:40:37 GMT -5
I'm going to give you some PROVISIONAL answers here..... (that means I'm guessing)...
Having Dirac run on all channels is a high priority so I expect that will be the case. I expect Dirac to initially run at 48k (but I cannot rule out the possibility that might change later). It is unlikely that any sort of hardware upgrade will be offered later specifically to increase the sample rate at which Dirac operates.
Keith, how about we stick to what the RMC-1 first ... seems like some basic questions are still on the table: 1. what is the sampling rate? 2. how many channels will be running at once at that rate. Why do I have a feeling that even 16 channels @ 48khz may be in question...
3. will you "enable" users to buy an upgrade module only for the RMC-1 to get back to what was originally promised for the sampling rate?
Keith this is one of the problems with Emotivas communication to it's customers. Instead of just guessing, you should make a joint opinion. Then you would find out if it was even possible and you could deside if it was a module you would make in the future and when you would make it, maybe not a specific date but instead something like after the speaker expansion modules is released and stabile. |
|
KeithL
Administrator
Posts: 10,517
|
Post by KeithL on Jan 29, 2020 14:45:20 GMT -5
In most situations where a single format is involved you start with the highest resolution and then create the others by down-sampling from there. I don't know offhand of anyone who actually makes simultaneous recordings of exactly the same content, in the same format, but at various sample rates. Sometimes, when a track is offered in both PCM and DSD, one is converted from the other, and sometimes separate recordings are made on different equipment. If the recording starts out in one format, then the question is not of which is better, but of which is closer to the original...
If one is converted from the other, then there is the distinct possibility that the conversion process has introduced slight differences. Even if you made recordings at different sample rates, using the same brand and model of recorder, there is still the fact that different sample rates will use different band-limiting filters, and so may sound different.
And, if they're recorded on different equipment, at the same time, then you have the possibility that the recorders themselves are imparting slightly different sonic signatures.
I still think your best bet is to start from a single high sample rate source file and, from it, produce two samples, at different sample rates, using a good quality sample rate converter. For example, start with a 24/192k PCM "original" - and, from that, make a 24/96k PCM copy and a 24/48k PCM copy, using the same conversion program, at its default settings. I think that, in practical terms, this is as close as it is possible for an end-user to find "two identical tracks at different sample rates".
Incidentally, here's a link to a rather comprehensive comparison of the performance of various sample rate converters...
Look at the various graphs provided for several different sample rate converters... Look at both the "Sweep" graphs and the others - each shows different performance characteristics.
You'll note that, while many are "arbitrarily perfect", many are quite far from ideal, and quite different from each other... (And some of the bad ones are in what we would otherwise consider to be "good programs".)
Note that, among others, both Audacity and r8Brain Free offer excellent performance - and are free to use.
And, in the category of paid products, Adobe Audition performs quite well.
And, obviously, if you don't notice a significant difference between 96k and 48k when playing music... Then it is highly unlikely that it will make an audible difference whether Dirac is run at one or the other. (And remember that things like differences in the design of the analog stages do make a significant difference in how gear sounds overall.)
The point, however, is that you cannot "just decide up front if your equipment and ears are going to miss the difference". Another name for "deciding things without any actual information" is "guessing"... I actually do plan on doing this with my own equipment. I'm in the process of building my streaming/ Hi-res player setup. I had a Marantz before that had a builtin hi-res player but the XMC-2 doesn't. I recently got Roon and I am going to setup a Raspberry Pi with Ropiee to stream and play back hi-res files. I think however to get a real answer you have to run multiple tests with files created from the source and not converted. Files created using the same hardware also. I don't know for sure how Hd-tracks creates their files do they go back to the source for each different resolution or do they just convert the file? Does the conversion "change" the sound in anyway? Lots of variables this is why I am for a actual scientific study. Create new files from the source with the same hardware. Well, just keep in mind that you won't have any say in the matter on the RMC-1 when you pump in a 96/24 file and it gets decimated. You will get what you get, so all you can do upfront is make up your own mind if your equipment and ears are going to miss the difference... if they are, then you should just assume that you'll miss it with Dirac in the absence of the ability to actually try it.
There was no difference in volume or any other artifacts I can detect in the conversion. I think some tracks you can buy from hdtracks, for example, at different sampling rates if you're worried about your conversion being the issue. Not sure if they will offer both 96/24 and 48/24 (they probably don't even go that low) but it's common to find 192/24 and 96/24 being offered.
|
|
KeithL
Administrator
Posts: 10,517
|
Post by KeithL on Jan 29, 2020 15:23:57 GMT -5
Basically yes.
What we're talking about here is a program running on a computer. The Dirac calibration program runs on your external computer... but the "correction engine" is a program running on the processor in the RMC-1.
The simple fact is that you can never know how fast a program will run, or how much processing power it will consume, until you actually have the program. At best, you can guess, based on what the programmers tell you, and on
previous experience with other more or less similar programs. In this case, the processors we chose to include in the RMC-1 are very powerful, but the code hadn't ever been written to run the Dirac correction engine on them before.
This was the main reason for all the delays...
apparently porting the code over to our processor proved far more
difficult for their programmers than Dirac had originally believed. (We simply assumed that, because the new processors are so much more powerful, we would "obviously" be able to achieve both higher speed and more channels...) I expect Dirac to initially run at 48k (but I cannot rule out the possibility that might change later). Question for you Keith...at the last Emofest (not the recent tech day, but at the last full Emofest), I recall Dan and Lonnie saying they built the RMC-1 to be powerful enough to run 96k, but what was ultimately be delivered was up to Dirac. Is that what's in play here as it relates to your comment? Thanks, Mark |
|
The griffin lites have about 4 times the processing power of the previous falcon chips that restricted chipset avrs and pre pro's to 7.1.4 I once read and they are dual core 
. Which then leads to the necessity of clarifying what is meant by "near future", and who's "future" and what's meant by "near". Maybe we all misunderstood and they meant "in the 'year' future." 
