bdiament
Junior Member
Rebel
Posts: 11
Audio Industry Affiliation: Producer, Engineer
|
Post by bdiament on Sept 12, 2019 3:25:29 GMT 10
Have seen previously that some consider wav files superior in SQ to flac files. Others have mentioned flac uncompressed is better due to metadata info. What have people here found best? Hi frednorc,
In my experience (and that of local listeners who participated in some tests we ran), FLAC always seems to change the sound of the source PCM file. Not necessarily a "night and day" change but not a perfect restoration either -- and this is with expansion occurring before listening rather than during listening (when we found the degradation to be greater).
I do all my recording, mastering, and personal listening to .aif format files. This format preserves metadata. To my ears, it is sonically indistinguishable from .wav format. (While some folks have told me they hear a difference, always in favor of .wav, concerted effort on my part has told me that as far as my ears are concerned, they are sonically identical.)
My best suggestion is to try some comparisons yourself to see if you hear any meaningful differences.
|
|
bdiament
Junior Member
Rebel
Posts: 11
Audio Industry Affiliation: Producer, Engineer
|
Post by bdiament on Sept 12, 2019 3:39:15 GMT 10
I find that we have to use the original resolution and format of the recording for best results. Even although we can mathematically change from 44.1 to 48K or even multiples, the result is garbage. Proper red book CD playback at 44.1 can be incredible. The world (except for Sony with patents) never needed 48K. I get great results from WAV, FLAC, AAC, OGG and raw files. I have no preference. Hi Robin,
I would submit that if the result of a sample rate conversion is garbage, it may well be the conversion algorithm that is to blame. After comparing dozens of such algorithms (and dither/noise shaping algorithms too), I've found that the preponderance tend to brighten and harden the sound. The effects are even more pronounced when the algorithms are used in real time (that is, while listening).
That said, there *are* one or two that allow conversion without--at least to my ears--exacting a sonic price. (And I still use these offline, that is, *not* while listening, saving the enjoyment for after the conversion is complete.) My criterion for evaluation is how close the converted file sounds to the unconverted original.
I know some engineers stick with integer conversion (even multiples, staying within a "family" of rates, such as 44.1k, 88.2, or 176.4k). With some algorithms, I can understand this, as they have less trouble with the "easier" math. However, the best ones I've heard create more transparent results, even with noninteger conversion than others do with the easier, integer conversion task.
I say interesting because I've found an unusually good correspondence between the measurements (particularly the 1k tone) and what I hear from the different algorithms.
I'm a big fan of Alexey Lukin's 64-bit SRC marketed by iZotope. (See the "steep, no alias" test with the 1k tone and compare with most others. While others show noise that "should be" inaudibly low in level, I've found that *something* is corresponding well with the degree of hardening I hear from those algorithms.)
This is the algorithm I use to convert my 192k recordings to 96k and to 44.1k. I've not (yet) found a way to make a CD sound more like the high res original.
|
|
sandyk (RIP Alex, 1939 - 2021)
Global Moderator
Posts: 226
About Me: Retired ex Principal Telecommunications Technical Officer with 43 years at Telstra (Australia)
I am also a Moderator in Hi Fi Critic Forum
Electronics hobbyist for >65 years with DIY projects including Loudspeakers, Stereo FM tuner, S/W Regen Receiver, Superhet AM ,
Synchrodyne PLL AM tuner (Phase Lock Loop),Stereo Tape Deck, Amplifiers including I.C. types, Class A, Class AB 100W/Ch. (ETI5000) 240W/Ch. Mosfet (AEM6000) ,several DACs , numerous PSUs including VERY low noise (<4uV) types etc.for myself and friends
Audio Industry Affiliation: NIL
|
Post by sandyk (RIP Alex, 1939 - 2021) on Sept 12, 2019 9:23:31 GMT 10
Have seen previously that some consider wav files superior in SQ to flac files. Others have mentioned flac uncompressed is better due to metadata info. What have people here found best? Hi frednorc,
In my experience (and that of local listeners who participated in some tests we ran), FLAC always seems to change the sound of the source PCM file. Not necessarily a "night and day" change but not a perfect restoration either -- and this is with expansion occurring before listening rather than during listening (when we found the degradation to be greater).
I do all my recording, mastering, and personal listening to .aif format files. This format preserves metadata. To my ears, it is sonically indistinguishable from .wav format. (While some folks have told me they hear a difference, always in favor of .wav, concerted effort on my part has told me that as far as my ears are concerned, they are sonically identical.)
My best suggestion is to try some comparisons yourself to see if you hear any meaningful differences.
Hi Barry
I have found that the quieter electrically the PC is, the less the difference between .flac and .wav/.aiff, but I have yet to get to the point where I am unable to hear a difference between them with good quality recordings. Wideband RF/EMI from Processor activity seems likely to be a major reason for this ?
Kind Regards Alex
|
|
|
Post by Audiophile Neuroscience on Sept 12, 2019 12:31:54 GMT 10
This is the algorithm I use to convert my 192k recordings to 96k and to 44.1k. I've not (yet) found a way to make a CD sound more like the high res original. Hi Barry, Most audiophiles find themselves with a lot of 16/44 material and the issue for some is whether upsampling (src, interpolation, whatever) improves SQ. In the case of 'upsampling' to DSD I suppose transcoding is also involved? Upsampling can be done by many DACs but also in software where a powerful computer is harnessed to do the heavy lifting. From my reading and limited understanding the theory usually revolves around moving aliasing into inaudible regions, avoiding PCM reconstruction filters in the case of upsampling to DSD, or other technical issues.
Technically it is beyond my pay grade but subjectively I have heard very expensive DACs sound worse to my ears when deploying their upsampling. The exception is my own Gryphon Kalliope DAC that has the option to upsample to 32bit 210 Khz and to which I keep on gravitating back to. This is despite my initial negative expectation bias. Why 210 KHz ? I have no idea, its certainly not a whole integer increase of 44.1 or 48. The DAC does have dedicated clocks for handling these families of Fs. So, like many things I am left pondering the empirical evidence. I suppose it really comes down to implementation at the end of the day. That and of course, Gryphon is magical......the head and wings of the eagle and body of the lion ! David ----------------------------------------------------------------------------------------------------------------------------- "All music is folk music. I ain't never heard no horse sing a song." - - Louis Armstrong
|
|
|
Post by ROWUK on Sept 13, 2019 4:28:54 GMT 10
In my case, 44.1/16 sound just fine at 44.1/16. 88.2/24 sounds just fine at 88.2/24 and 176.4/24 sounds best at this resolution. I have never had transcoding give me equal results to the original but Barrys suggestion of iZotope is one that I will try. I can always learn something.
|
|
bdiament
Junior Member
Rebel
Posts: 11
Audio Industry Affiliation: Producer, Engineer
|
Post by bdiament on Sept 13, 2019 21:03:57 GMT 10
Hi Barry
I have found that the quieter electrically the PC is, the less the difference between .flac and .wav/.aiff, but I have yet to get to the point where I am unable to hear a difference between them with good quality recordings. Wideband RF/EMI from Processor activity seems likely to be a major reason for this ?
Kind Regards Alex
Hi Alex, Differences were consistently noted here even with the laptop running on battery power. Again, not what I'd call "night and day" but nothing like identical either-- not hard to differentiate. (Personally, unless one is *really* pressed for space, I don't see the point.)
|
|
bdiament
Junior Member
Rebel
Posts: 11
Audio Industry Affiliation: Producer, Engineer
|
Post by bdiament on Sept 13, 2019 21:10:57 GMT 10
This is the algorithm I use to convert my 192k recordings to 96k and to 44.1k. I've not (yet) found a way to make a CD sound more like the high res original. Hi Barry, Most audiophiles find themselves with a lot of 16/44 material and the issue for some is whether upsampling (src, interpolation, whatever) improves SQ. In the case of 'upsampling' to DSD I suppose transcoding is also involved? Upsampling can be done by many DACs but also in software where a powerful computer is harnessed to do the heavy lifting. From my reading and limited understanding the theory usually revolves around moving aliasing into inaudible regions, avoiding PCM reconstruction filters in the case of upsampling to DSD, or other technical issues.
Technically it is beyond my pay grade but subjectively I have heard very expensive DACs sound worse to my ears when deploying their upsampling. The exception is my own Gryphon Kalliope DAC that has the option to upsample to 32bit 210 Khz and to which I keep on gravitating back to. This is despite my initial negative expectation bias. Why 210 KHz ? I have no idea, its certainly not a whole integer increase of 44.1 or 48. The DAC does have dedicated clocks for handling these families of Fs. So, like many things I am left pondering the empirical evidence. I suppose it really comes down to implementation at the end of the day. That and of course, Gryphon is magical......the head and wings of the eagle and body of the lion ! David ----------------------------------------------------------------------------------------------------------------------------- "All music is folk music. I ain't never heard no horse sing a song." - - Louis Armstrong Hi David, I believe there *can* be some advantage to using a good sample rate conversion algorithm to resample in an upward direction. This, if the SRC is done offline (that is, as a separate process, before listening, as opposed to online (during listening) AND if the DAC's filters at 1x rates (44.1) are truly inferior to the same DAC's filters at higher rates. That said, my experience has been that many DACs just don't do the clocking particularly well at the higher rates (which I believe accounts for some folks finding the higher rates "inferior"). In practice, I find that with a well-designed DAC, I'd rather just listen at the original rate since all that is added is the higher rate and *not* more information. With a good design, I don't find anything gained by adding another process.
|
|
bdiament
Junior Member
Rebel
Posts: 11
Audio Industry Affiliation: Producer, Engineer
|
Post by bdiament on Sept 13, 2019 21:18:03 GMT 10
In my case, 44.1/16 sound just fine at 44.1/16. 88.2/24 sounds just fine at 88.2/24 and 176.4/24 sounds best at this resolution. I have never had transcoding give me equal results to the original but Barrys suggestion of iZotope is one that I will try. I can always learn something. Hi Robin, I would not say that the results equal the original, just that the usual degradations of most sample rate conversion algorithms are not necessarily endemic to converting the sample rate. To my ears, the better algorithms simply convert the rate. The inherent differences between the rates still exist: 44.1 just isn't going to sound like 96 (or 88.2) and neither will sound like 192 (or 176.4). And for upward conversion, while no information is added, the DAC's filter may have an easier job but its clocking comes into play, which might be the biggest concern with most DACs I've heard. Personally, I listen to 44.1 at 44.1. It is what it is. ;-} (As an aside, when I need to convert, such as taking a 192 original and making a 44.1 version for CD, iZotope's 64-bit SRC with the right settings seems to "get out of the way" better than the few dozen others I've tried. I would say the same for their MBIT+ dither/noise shaping algorithm. Both are the work of a young (compared to me) programmer named Alexey Lukin.
|
|
sandyk (RIP Alex, 1939 - 2021)
Global Moderator
Posts: 226
About Me: Retired ex Principal Telecommunications Technical Officer with 43 years at Telstra (Australia)
I am also a Moderator in Hi Fi Critic Forum
Electronics hobbyist for >65 years with DIY projects including Loudspeakers, Stereo FM tuner, S/W Regen Receiver, Superhet AM ,
Synchrodyne PLL AM tuner (Phase Lock Loop),Stereo Tape Deck, Amplifiers including I.C. types, Class A, Class AB 100W/Ch. (ETI5000) 240W/Ch. Mosfet (AEM6000) ,several DACs , numerous PSUs including VERY low noise (<4uV) types etc.for myself and friends
Audio Industry Affiliation: NIL
|
Post by sandyk (RIP Alex, 1939 - 2021) on Sept 13, 2019 21:37:09 GMT 10
Hi Barry I found that my highly modified Musical Fidelity X-DAC V3 that I use with my PC, sounded quite noticeably better with high resolution files after changing the Xtal Oscillator to a .1PPM 24.576 MHZ TCXO type (Temperature Controlled Xtal Oscillator) with a lower noise power supply, but it didn't make much difference for 44.1 though.
Kind Regards Alex
|
|
|
Post by ROWUK on Sept 14, 2019 1:11:21 GMT 10
In my case, 44.1/16 sound just fine at 44.1/16. 88.2/24 sounds just fine at 88.2/24 and 176.4/24 sounds best at this resolution. I have never had transcoding give me equal results to the original but Barrys suggestion of iZotope is one that I will try. I can always learn something. Hi Robin, I would not say that the results equal the original, just that the usual degradations of most sample rate conversion algorithms are not necessarily endemic to converting the sample rate. To my ears, the better algorithms simply convert the rate. The inherent differences between the rates still exist: 44.1 just isn't going to sound like 96 (or 88.2) and neither will sound like 192 (or 176.4). And for upward conversion, while no information is added, the DAC's filter may have an easier job but its clocking comes into play, which might be the biggest concern with most DACs I've heard. Personally, I listen to 44.1 at 44.1. It is what it is. ;-} (As an aside, when I need to convert, such as taking a 192 original and making a 44.1 version for CD, iZotope's 64-bit SRC with the right settings seems to "get out of the way" better than the few dozen others I've tried. I would say the same for their MBIT+ dither/noise shaping algorithm. Both are the work of a young (compared to me) programmer named Alexey Lukin. Then it probably makes sense to just stick with the current Focusrite ADC/DACs and original resolution for home and try iZotope for recording productions.
|
|