Changing the bitrate and sampling rate
Posted by: Hot Rats on 26 July 2010
I was reading a post on a board in which someone was extolling the virtues of changing a file ripped from a CD (16-Bit/44.1kHz) to either 24-Bit/96kHz or 24-Bit/192kHz for replay through a DAC.
I can't see it myself as no additional information is gained from doing this but is there anything to be gained by doing this?
I can't see it myself as no additional information is gained from doing this but is there anything to be gained by doing this?
It's called upsampling.
Theoretically this is a very good approach. Many DACs employ run-time upsampling -- from cheap PS Audio DLink III and Cambridge Audio (700$) up to the most expensive like for example Scarlatti (80,000 $).
Regarding new information -- of course upsampling does not increase it. But if you read engineers induction into digital audio (for example Dan Lavry) you will see a very serious foundation for 16/44 format to cover whatever a man can capture with his ears. Many well-known designers share this opinion. From this perspective going into hi-res creates the advantage not from increasing the amount of information but rather from simplifying the DAC's efforts. In other words it's easier to design 24/96 DAC to make it sing with hi-res than to design a 16/44 DAC to sing the same good with 16/44 -- although a man's ear definitely truncates both to what 16/44 is theoretically capable of. Many designers oppose this approach and prefer design "native" D/A conversion without upsampling.
But there is immediate effect from pre-DAC digital-digital processing: the upsampled data are processed in hi-res mode of the DAC chip -- which is (theoretically) better (smoother) than in 16/44 mode. It contains less grain and sounds closer to vinyl. You can check this effect easily on your nDAC -- I was really impressed on how smoother nDAC plays on hi-res rather than 16/44. Similar effect with KDS.
Of course it depends on implementation -- don't expect that cheapish upsampling DACs will outperform Red Book CD555. BB-1704 in hi-res mode requires extremely low jitter to show the benefits of upsampling. It also requires high order filtering on the output. Many people prefer CD555 to the ultramodern jitter(-free) nDAC design
What I found with upsampling DACs is they are mostly "unnatural". They sound very cold and extra "airy" as opposite to natural and warm vinyl presentation. Please, don't extend this opinion to the mighty upsampling DACs which I did not even audition.
One of explanations of this "cold airy sound" is that simple upsampling process usually creates pre-echo of the original step-wise waveform -- like a small echo preceding the waveform front. This is exactly the area where apodizing filters can be used to prevent form pre-echo. The other explanation is the upsampling creates ultrasound contamination which is reproduced through increased frequency range of hi-res sampling. Many people by the way like this cold upsampled sound and consider it as a hi-fi quality (not me).
Anyway -- I believe one can build a software which will process a "clever" upsampling so that playing the upsampled file through hi-res DAC will give much better results than playing the original 16/44 file through the same DAC.
I also believe that once jitter is almost resolved nowdays there will be a come back to "native" converters and they can be a big rival to upsampling->hi-res converters.
I think the difference between two approaches will be not that big and competition will be most probably based on market indicators -- rather than sonic terms. For example, if hi-res files will evolve around in abundance -- then hi-res DACs will get a priority and upsampling of old 16/44 will be the easiest way to go even if it gives up to "native" DACs.
Theoretically this is a very good approach. Many DACs employ run-time upsampling -- from cheap PS Audio DLink III and Cambridge Audio (700$) up to the most expensive like for example Scarlatti (80,000 $).
Regarding new information -- of course upsampling does not increase it. But if you read engineers induction into digital audio (for example Dan Lavry) you will see a very serious foundation for 16/44 format to cover whatever a man can capture with his ears. Many well-known designers share this opinion. From this perspective going into hi-res creates the advantage not from increasing the amount of information but rather from simplifying the DAC's efforts. In other words it's easier to design 24/96 DAC to make it sing with hi-res than to design a 16/44 DAC to sing the same good with 16/44 -- although a man's ear definitely truncates both to what 16/44 is theoretically capable of. Many designers oppose this approach and prefer design "native" D/A conversion without upsampling.
But there is immediate effect from pre-DAC digital-digital processing: the upsampled data are processed in hi-res mode of the DAC chip -- which is (theoretically) better (smoother) than in 16/44 mode. It contains less grain and sounds closer to vinyl. You can check this effect easily on your nDAC -- I was really impressed on how smoother nDAC plays on hi-res rather than 16/44. Similar effect with KDS.
Of course it depends on implementation -- don't expect that cheapish upsampling DACs will outperform Red Book CD555. BB-1704 in hi-res mode requires extremely low jitter to show the benefits of upsampling. It also requires high order filtering on the output. Many people prefer CD555 to the ultramodern jitter(-free) nDAC design
What I found with upsampling DACs is they are mostly "unnatural". They sound very cold and extra "airy" as opposite to natural and warm vinyl presentation. Please, don't extend this opinion to the mighty upsampling DACs which I did not even audition.
One of explanations of this "cold airy sound" is that simple upsampling process usually creates pre-echo of the original step-wise waveform -- like a small echo preceding the waveform front. This is exactly the area where apodizing filters can be used to prevent form pre-echo. The other explanation is the upsampling creates ultrasound contamination which is reproduced through increased frequency range of hi-res sampling. Many people by the way like this cold upsampled sound and consider it as a hi-fi quality (not me).
Anyway -- I believe one can build a software which will process a "clever" upsampling so that playing the upsampled file through hi-res DAC will give much better results than playing the original 16/44 file through the same DAC.
I also believe that once jitter is almost resolved nowdays there will be a come back to "native" converters and they can be a big rival to upsampling->hi-res converters.
I think the difference between two approaches will be not that big and competition will be most probably based on market indicators -- rather than sonic terms. For example, if hi-res files will evolve around in abundance -- then hi-res DACs will get a priority and upsampling of old 16/44 will be the easiest way to go even if it gives up to "native" DACs.