Oversampling CDT + Upsampling DAC ???

[esperowl]

Gents, I have an Audionote DAC that is non-filtering and upsamples x1.
I've been looking for a dedicated transport to partner it with and have been offered a CEC transport.
The CEC both filters and oversamples AFAIK and I was wondering about the compatibility of these units.
Good idea? Makes no difference? Disastrous?

[Yomanze]

The CDT's digital out won't be doing any upsampling or oversampling, you're probably talking about the CEC's internal DAC, which wouldn't be used for transport duties.

[nat8808]

And isn't "1x upsampling" another way of saying there is NO upsampling?

I thought Audio Note digital products were famous for no digital filters, no oversampling, no upsampling ?

[nat8808]

Yep, the "1xUpsampling" is nothing but a marketing term to mean it does nothing.


Therefore philosophically there is something wrong with digital filtering and this is proven in practical listening tests. Hence we do not oversample the input signal or digitally tamper with it at all. Hence 1x oversampling (not oversampled). And this also precludes the use of bitstream and delta sigma DACs, which rely on processing. Therefore we only use resistor ladder type DACs.

[esperowl]

The CDT's digital out won't be doing any upsampling or oversampling, you're probably talking about the CEC's internal DAC, which wouldn't be used for transport duties.

Doh! This is why I need hifi forums - so that the bleedin' obvious can be pointed out to me. Many thanks!

[esperowl]

Yep, the "1xUpsampling" is nothing but a marketing term to mean it does nothing.

"x1 upsampling" is emblazoned on the front of my DAC. I'm surprised AN deemed it necessary to do that given that their own blurb says that... oh hold on, that's not the same as 1x oversampling though is it.

[Yomanze]

Upsampling is increased bit depth e.g. 16bit to 24bit with the goal of increasing dynamic range. Oversampling is running a file at multiples of its original frequency in order to allow for noise shaping and gentler filter slopes, but it can also be used by DACs to achieve resolution (work in the first place!) e.g. 1 bit, massively oversampling, bitstream DACs. Both are audible and represent some of the biggest differences in DACs. I have had some interesting experiences with software oversampling (from 44.1kHz to 88.2kHz without upsampling from 16bit) recently with my NOS DAC, which corrects the treble rolloff at the expense of more digital artefacts. 96kHz oversampling sounded less natural to 88.2kHz likely due to the immensely more complex calculations (44.1kHz x 2 is easier). I have also been listening between linear and minimum phase digital filtering via software, interesting to do these without resorting to hardware (like most modern DACs have built into the chip).

[nat8808]

Upsampling and oversampling are only about sampling frequency (hence the "sampling" in both), not about changing bit depth.

AND, they are essentially the very same thing from a technical standpoint.

Oversampling is a multiplying of the sampling frequency by an integer, normally a power of two. This often happens as part of the D/A process, integral to the digital filter, part of the design of the whole DAC really. I'm a bit hazy on digital sampling theory but I think oversampling was devised to better implement digital filters to provide the filtering required by the nyquist sampling theorum .

Upsampling is a word from the studio world really where instead of just multiplying by an integer for a technical purpose, you are actively choosing the final higher desired sampling rate. e.g. if you have music from a CD (44.1KHz) and you want to use it for standard digital video (normally 48KHz) you upsample from 44.1 to 48kHz. Upsampling is a general term to encompase moving up from any sample rate to a higher one. When it's not a straight multiple like from 44.1 to 96KHz, then the maths is a bit rough and ready and you run in to numbers with more decimal places than the DSP can handle so there are complicated mathematical algorithms to make the conversion as good as possible. Different companies will use different algos (though, these days I bet there are some often used standard ones) and some like Anagram/Edel make sales on their proprietory upsampling techniques.

But, when choosing to upsample from 44.1 to 172.4, it's no different from 8 x oversampling.

Many chips will ALWAYS oversample, so when you upsample from say 44.1 to 192KHz, internally the DAC chip will be also oversampling and converting at rates into the megahertz.

I make no claims that any of the above is well explained nor makes sense.

[nat8808]

"x1 upsampling" is emblazoned on the front of my DAC. I'm surprised AN deemed it necessary to do that given that their own blurb says that... oh hold on, that's not the same as 1x oversampling though is it.

They are the same thing - it means the digital signal goes in and doesn't undergo any changing of sample rate, either before or during the conversion process. It has always been AN's philosophy that they don't do oversampling so I guess they've never put that on their products. Upsampling became a hifi catchword some time ago, so they've bothered to put that on the front just as a reminder that it's different from all the other upsampling-hyped DACs from the time.

[Yomanze]

Upsampling and oversampling are only about sampling frequency (hence the "sampling" in both), not about changing bit depth.

AND, they are essentially the very same thing from a technical standpoint.

Oversampling is a multiplying of the sampling frequency by an integer, normally a power of two. This often happens as part of the D/A process, integral to the digital filter, part of the design of the whole DAC really. I'm a bit hazy on digital sampling theory but I think oversampling was devised to better implement digital filters to provide the filtering required by the nyquist sampling theorum .

Upsampling is a word from the studio world really where instead of just multiplying by an integer for a technical purpose, you are actively choosing the final higher desired sampling rate. e.g. if you have music from a CD (44.1KHz) and you want to use it for standard digital video (normally 48KHz) you upsample from 44.1 to 48kHz. Upsampling is a general term to encompase moving up from any sample rate to a higher one. When it's not a straight multiple like from 44.1 to 96KHz, then the maths is a bit rough and ready and you run in to numbers with more decimal places than the DSP can handle so there are complicated mathematical algorithms to make the conversion as good as possible. Different companies will use different algos (though, these days I bet there are some often used standard ones) and some like Anagram/Edel make sales on their proprietory upsampling techniques.

But, when choosing to upsample from 44.1 to 172.4, it's no different from 8 x oversampling.

Many chips will ALWAYS oversample, so when you upsample from say 44.1 to 192KHz, internally the DAC chip will be also oversampling and converting at rates into the megahertz.

I make no claims that any of the above is well explained nor makes sense.

Makes sense, I get confused between the terminology & stand corrected! Indeed oversampling was originally conceived because the 14 bit CD specification was changed to 16 bit, and Philips had to figure out a way to get 16 bit resolution from an already built 14 bit chip, so the answer was oversampling.

[nat8808]

Upsampling is increased bit depth e.g. 16bit to 24bit with the goal of increasing dynamic range. Oversampling is running a file at multiples of its original frequency in order to allow for noise shaping and gentler filter slopes, but it can also be used by DACs to achieve resolution (work in the first place!) e.g. 1 bit, massively oversampling, bitstream DACs. Both are audible and represent some of the biggest differences in DACs. I have had some interesting experiences with software oversampling (from 44.1kHz to 88.2kHz without upsampling from 16bit) recently with my NOS DAC, which corrects the treble rolloff at the expense of more digital artefacts. 96kHz oversampling sounded less natural to 88.2kHz likely due to the immensely more complex calculations (44.1kHz x 2 is easier). I have also been listening between linear and minimum phase digital filtering via software, interesting to do these without resorting to hardware (like most modern DACs have built into the chip).

Changing bit depth is nothing more than essentially tagging the file as either 16bit or 20bit etc and making the file the apropriate data length too by filling it with empty information i.e. the sonic information in the file stays exactly the same and the sonis data still 16bit but it 'looks' like the higher rate to the equipment which doesn't know the difference between any old numbers or sonic data anyway.

Noise shaping and dithering is all about ways of recording something at, for example, 20bit and putting it on a 16bit format like CD and still then getting near 20bit effective resolution on playback via a minimum 20bit DAC. Basically it applies more bit depth to frequency ranges at which our ears are more able to discern the detail and so it sounds like, say 20 bit to the human ear.

Normally noise shaping is done at the recording end but I think the buzz of upsampling also has ended up with claims of real sonic benefits from increasing the bit depth of a fixed bit depth file via noise shaping/dither. It's become standard to do both now, upsample and increase bit depth.