fr0g said:
If the 'studio master' was encoded at 192kbs, then surely you were upconverting to CD quality, not downsampling, in which case I wouldn't expect there to be any differences at all.
If the 'studio master' was encoded at 192kbs, then surely you were upconverting to CD quality, not downsampling, in which case I wouldn't expect there to be any differences at all.
spiny norman said:
This is what fr0g actually said:
"Here is an MP3 in low bitrate of the Linn 24/192 test track"
He then compared that to another file encoded at CD quality from the original , so I think you misunderstood.
Ah right, I see, I think.
I'm afraid he confused me so much with his comments about testing and not testing MP3 files and what he was actually trying to do that I may have made an error. Apologies.
I'm afraid he confused me so much with his comments about testing and not testing MP3 files and what he was actually trying to do that I may have made an error. Apologies.
spiny norman said:
Just to be completely clear.
I downloaded the HD track from Linn.
This was a 24 bit 192 KHz sample.
I converted to 16 bits and 44.1 KHz using Audacity.
I then inverted one of the tracks
I then combined the 2 tracks which means anything that is the same in both samples is now removed.
So any differences are now left over.
There were differences... Some noise at 17-20 KHz, at -84 dB (ie inaudible) and some more noise at 80 KHz or so, at -70dB (also inaudible)
The file DIFFERENCE.FLAC is the result. This is a 16 bit, 192 KHz file.
If you play it back you are likely not to hear anything.
That is the difference between a well mastered 24/192 file, and the same one downsampled to CD quality...
And for the benefit of those that still don't understand about why I don't do the test on MP3.
The MP3 codec is lossy..so it does lose audio information. The magic is in "which" information it decides to lose.
When you play back any track, there will be parts of the audio that are inaudible, "because" they are masked by louder sounds. This is one of the tricks that MP3 (or any lossy codec) uses. The lower the bitrate, the more aggressive this gets, and why when you get to 128 Kbps and lower you start to hear a reduction in quality.
However, testing for "differences" in this instance WILL give you a track that you can hear. We know that. It proves nothing about its audibility in the real world, ie playing back the actual song. The only way we can test that is by ABX, which everyone loves!
On 2 lossless tracks, we can do that test, as those near silent audio samples are retained.
steve_1979
Well-known member
spiny norman said:
That looks like a typo. I'd guess it was supposted to be 192 kHz.
As in a 24/192 studio master that's downsampled to 16/44.1 CD quality.
I've just caught up with this thread after a busy weekend. A lot of good contributions and links and my thanks to everyone that have contributed in a positive way.
I genuinely feel that my understanding has improved and that, for me, is a good thing. However there are still some questions remaining.
Firstly, we know that a signel needs to be filtered prior to processing, to remove any signels beyond half the sampling rate, and we know that these filters can be of different types, analogue, digital or quite possibly a combination of the two. We also know that the filters used in playback dacs can be shown to measre differently and that some listeners find these differences quite audible. We have a fair number of dacs that offer these different filters as a user selectable option.
If the filters do change the percieved results in an audible way, what has happened to the idea of 'perfect' or 'transparent' playback?
Secondly, it has been explained how the use of hi-res files in the recording process provide 'digital headroom' for the manipulations of the signal that takes place and I sort of get that. However, Oldric has pointed out that the inclusion of DSP in the replay chain is manipulating the 16 bit signal without any headroom so that the manipulation will effect the 16 bits.
I suspect that, as with digital volume, 'extra bits' will be added and the processing carried out at higher bit rates but it would be nice to know in perhaps a little more detail.
I genuinely feel that my understanding has improved and that, for me, is a good thing. However there are still some questions remaining.
Firstly, we know that a signel needs to be filtered prior to processing, to remove any signels beyond half the sampling rate, and we know that these filters can be of different types, analogue, digital or quite possibly a combination of the two. We also know that the filters used in playback dacs can be shown to measre differently and that some listeners find these differences quite audible. We have a fair number of dacs that offer these different filters as a user selectable option.
If the filters do change the percieved results in an audible way, what has happened to the idea of 'perfect' or 'transparent' playback?
Secondly, it has been explained how the use of hi-res files in the recording process provide 'digital headroom' for the manipulations of the signal that takes place and I sort of get that. However, Oldric has pointed out that the inclusion of DSP in the replay chain is manipulating the 16 bit signal without any headroom so that the manipulation will effect the 16 bits.
I suspect that, as with digital volume, 'extra bits' will be added and the processing carried out at higher bit rates but it would be nice to know in perhaps a little more detail.
Thanks (I think!) for the clarification, fr0g, but the alternative 'clarifications' offered by others seems to suggest that at least some of those debating here aren't quite sure what they're talking about. Which is worrying...
davedotco said:
this issue has already been picked up a couple of times many pages ago by Matt49, Shadders and my-humble-self. but no one else seemed to have grasped the problem. seems like followers of 16/44.1 can see only the theoretical side of things and can't see into the actual D2A process that's happening in DACs. nice you've noticed this aspect of digital playback too :cheer:
davedotco said:
this is quite correct too. like I said in my earlier post legacy hi-fi is propably the last bastion of analog sound processing. in which case 16 bits offer as much headroom as you'd need because the digital signal gets converted into analog without eny DSP applied to it. however, this is 21st century now. DSP engines are efficient, small and cheap so can be easily implemented in replay chain giving a lot of advantage in signal manipulation over analog processing, in order to overall enhance the sound quality of the system. in fact most, if not all, hi-quality speaker systems on the market, like from B&W, Audio Pro, B&O, Geneva to name but few, already use DSP too fine tune how their speakers should actully sound like. this is greatly beneficial. just imagine how would a Zeppelin sound like if it was a passive speaker system with no equalisation... anyway, as you propably knew DSP can be used to sort out many things, like volume control (BTW no analog volume control will ever come close to the precission of adjusting volume in digital domain), xovers in active speakers, in-phase integration of drives outputs, compensting for less-than-perfect impulse responses of raw output of the drivers, compensating for non linear response of the drivers, etc. etc.
in this light I'd be inclined to say that 24 bits offer better use for playback purposes then 16 bits, provided you opt for a DSP enhanced system. this especially holds true in case of volume control. 8 bits of resolution more over 16 bits offers you some 50dB of attenuation. I bet you won't need more. and since studios already record in 24 bits why opt for clinging to Red Book? I understand 24bit music is now sold at a premium and that DSP empowered speaker system are yet not a common place in the hi-fi world but everything will change in time IMO. music in 24 bit resolution will become a standard and more and more manufacturers will be using DSP to start offering better and better performing products.
I noticed when buying some music from LinnMusic the other day that many "Studio master" downloads were in fact 24/44.1
I would have no problem if 24 bit became a standard, although I would end up having to downsample to 16 bits each time. Whilst storage is cheap, bandwidth is still an issue.
A 16/44 album in FLAC is typically 250 MB
A 24/44.1 is typically over 500 MB
A 24/96 is typically well over a Gigabyte
and 24/192 is typically well over 2 GB. (So in effect you are downloading 1.5 GB of silence)
From my own tests it's apparent that this extra space is not really needed.
I would have no problem if 24 bit became a standard, although I would end up having to downsample to 16 bits each time. Whilst storage is cheap, bandwidth is still an issue.
A 16/44 album in FLAC is typically 250 MB
A 24/44.1 is typically over 500 MB
A 24/96 is typically well over a Gigabyte
and 24/192 is typically well over 2 GB. (So in effect you are downloading 1.5 GB of silence)
From my own tests it's apparent that this extra space is not really needed.
Thanks Oldric. Some more thoughts though.......
Digital volume is widely used in dacs and players for 16 bit playback. Noise/extra bits are added for processing at (usually) 24 bits, giving the headroom required for the 48 dB level reduction you speak of. Would something similar provide the headroom for the DSP to operate?
The filter issue, I find a bit bizarre. I have heard digital processors demonstrated in such a way that they are completely transparent as long ago as the mid 80s, with the PCM F1 configured back to back. Yet we also know that different filters measure and sound different, so how do we get the processing to be 'audibly transparent' when compared to the original. If we can hear the differences between filters then surely the filters should alter the sound compared to the original in some way.......... :?
Digital volume is widely used in dacs and players for 16 bit playback. Noise/extra bits are added for processing at (usually) 24 bits, giving the headroom required for the 48 dB level reduction you speak of. Would something similar provide the headroom for the DSP to operate?
The filter issue, I find a bit bizarre. I have heard digital processors demonstrated in such a way that they are completely transparent as long ago as the mid 80s, with the PCM F1 configured back to back. Yet we also know that different filters measure and sound different, so how do we get the processing to be 'audibly transparent' when compared to the original. If we can hear the differences between filters then surely the filters should alter the sound compared to the original in some way.......... :?
fr0g said:
Totally agree fr0g.
If the replay chain is a straightforward replay system then I am happy with 16 bits.
How do you react to the thoughts in recent posts that, as dsp in the playback chain becomes more commonplace, we need the extra bits to provide the 'digital headroom' so that the manipulation of the signal does not impinge on the 16 bit (music) signal?
For 'purist' hi-fi use I would want to think that the dsp does not effect the 16 bits that we have decided is the standard required for accurate music reproduction.
I am not even sure if what I am asking makes sense, hence the need for clarification.
davedotco said:
To be honest Dave, I don't really react to the DSP question...probably as it doesn't affect me
Like I said, I don't object to 24 bit becoming standard (so long as I retain the ability to download the equally good quality 16 bit version). I just object to the dishonesty of it all and subsequent high pricing.
John Duncan
Well-known member
shadders
Well-known member
fr0g said:To be honest Dave, I don't really react to the DSP question...probably as it doesn't affect me
Like I said, I don't object to 24 bit becoming standard (so long as I retain the ability to download the equally good quality 16 bit version). I just object to the dishonesty of it all and subsequent high pricing.
Hi,
Yes - Hifi News analyse the downloads and previously many are resampled 44.1kHz or 48kHz.
In March 2014 issue - one 96kHz 24bit FLAC recording was filtered at 26kHz.
Recently they all seem to be ok - but with odd spurious frequency issues (especially if DSD resampled).
Hifi News provide a very good service here.
I have bought 4 x Pure Audio Blu-ray discs - their premise is that they will provide 96kHz 24bit - no resampling - i await to see any anaylsis of these.
Regards,
Shadders.
I don't think it much matters what it is called.
It is the deliberate 'hobbling' of decent recordings to pander to the mass market that really matters. As Clare pointed some time ago the record industry is about making money and while it continues to believe that highly compressed, mastered as loud as possible, releases will be more successful, they will continue to produce just that.
My own reaction to that, is as I have said before, to consider such releases as 'trash', not to buy it and to ignore it in every way possible. There are quite literally thousands upon thousands of decent recordings of good music that I have yet to hear, I shall spend my time and money on them.
It is the deliberate 'hobbling' of decent recordings to pander to the mass market that really matters. As Clare pointed some time ago the record industry is about making money and while it continues to believe that highly compressed, mastered as loud as possible, releases will be more successful, they will continue to produce just that.
My own reaction to that, is as I have said before, to consider such releases as 'trash', not to buy it and to ignore it in every way possible. There are quite literally thousands upon thousands of decent recordings of good music that I have yet to hear, I shall spend my time and money on them.
oldric_naubhoff said:
I explained a few pages back that DSP engines need more bits of resolution than their inputs to avoid truncation. Just because a DSP engine is 24 or 32 bits does not imply the input needs to be 24 or 32 bits - the exact opposite in fact. It is expected that a 16 bit signal will need 24+ bits of DSP resolution to avoid truncation and distortion.
andyjm said:
Excellent, so the 'headroom' is provided by the dsp not by 'extra bits' in the incoming data.
Any thoughts on the question of filters that I raised a liitle earlier?
shadders
Well-known member
davedotco said:
Hi,
The filter issue - i am unsure as to what you are referring to.
For the PCM F1 - you have stated that this is a processor - so it may not be filtering, perhaps analysing the data as per some network analysers - do not change the data - but can provide information regarding the data stream ?
A filter as its name implies - will filter out something, and pass something else. Such as filtering out the high frequency content and passing the low frequency content.
Depending upon the filter design, the filter may have in band ripple - such that the attentuation in band is not a straight line.
I am sure that digital filters exist (please can someone confirm) that have no in band ripple - but essentially, a "digital" filter by its very purpose will modify the data stream - the audio data - before it is converted to an analogue signal.
Is this sufficent explanantion - or have i misunderstood your question on filters ?
Regards,
Shadders.
shadders said:
so in order to avoid the necessity of using a digital anti-aliasing filter we would have to sample the analog input signal, as picked up by the microphones, at high enough sampling frequency so that the Nyquist frequency lies above the highest frequency being sampled. am I right?
shadders
Well-known member
oldric_naubhoff said:
Hi,
I recall that the theory is that the analogue input signal to the ADC is band limited such that the highest frequency appearing in the analogue signal to the ADC is at most 1/2 the sampling frequency. That is the input signal is analogue filtered before sampling.
Since a brick wall filter in the analogue domain is not possible - the filter is usually a high order whose cut off frequency (-3dB point) is less than the 1/2 x sampling frequency.
In the reconstruction of the analogue signal - DAC - again, the output analogue filter will remove all components (as much as possible) that are greater than 1/2 x sampling frequency.
Digital filters are used in a DAC when they oversample the data stream - to interpolate the analogue data at the higher sample rate. There is usually an analogue filter on the output of the DAC - but is generally lower in order - less components, to remove the image frequencies (anti aliasing filter).
Regards,
Shadders.
matt49
Well-known member
oldric_naubhoff said:
Yes, this is correct. One reason for 24 bit recording is to provide that headroom. (A couple of us I went into this in detail many pages back.) But it's not necessary at the playback stage.
On the question of 24-bit playback systems, there are various reasons why 24-bit can make sense, e.g. using 8 bits for digital attenuation as Sonos does, but none of this has anything to do with the intrinsic sound quality of the 16-bit digital stream.
Matt
shadders said:
I am primarily trying to understand how different filters apply in the dac. We have seen dacs come to market that have a choice of filters selectable by the user. These measure differently and sound different, so which one is 'right'.
If, as seems likely, all filters have an effect on the sound, how can the dac be 'transparent'.
The PCM F1 was a stand alone device that provided two channels of a to d conversion and two channels of d to a conversion, all at 16/44.1. It could be set up so that the digital output can be fed into the digital input, providing a full a to d to a process.
In one demonstration I took part in, 4 F1 units were used back to back to provide 8 processors and 4 complete a to d to a cycles. It was switched in and out of an otherwise analogue system and no one could reliably tell the difference. Bare in mind that this was a processor designed and built in the early 80s.
Facebook
Twitter
Instagram