[gregorio]

When converting from 24 bit to 16 bit, there may be truncation or omission of low-level bits of information from the digital audio data.

Again, we do not convert from 24bit to 16bit, where did you get that? Studio mixes and masters are typically 64bit float (or previously 56bit fixed). In the case of converting that to 16bit, noise-shaped dither is applied so that what is truncated is below -120dB.

if any sound information is truncated or missing, it means that the resolution has decreased.

In theory yes but in practice, not in the slightest. The only information below -120dB is noise and distortion. In fact, as almost no recordings have a dynamic range greater than about 60dB, then pretty much everything below -60dB is just noise/distortion, let alone 1,000 times lower level than that! And, noise/distortion below-120dB is nowhere even close to being audible.

To be sure, the converted 16-bit file is already different from the original 24-bit studio-master file, and this is not something the most demanding audiophile wants for the highest sound quality.

A 16bit conversion is obviously different from a 24bit version, just as a 24bit version is obviously different from the 64bit mix/master because they obviously have a fewer number of bits per sample. However, that makes absolutely no difference to sound quality because the downstream chain, DAC + amp + transducers cannot reproduce the sound of that noise/distortion, as it’s well below the noise floor of that chain.

If you can't hear the difference in sound between 16-bit 44.1KHz vs 24-bit 96KHz studio-master recordings, this doesn't mean all humans can't hear the difference in sound either.

It’s obviously got nothing to do with what I can hear. As what we’re removing (below -120dB) is not even reproducible by the downstream chain, how can anyone hear what’s not even in the sound coming out of their transducers? And even if it were in the sound, it would be well below human audibility anyway.

Also, humans can’t hear above 20kHz and adults typically lower than 16kHz. So it DOES mean all humans can’t hear the difference!

And don't forget to mention, DSD still has a higher resolution than CD

Do you really believe 1bit is higher resolution than 16bit? Obviously it’s far lower resolution but with noise-shaped dither it’s perceptually about the same.

G

 

[CiptoH]

Do you really believe 1bit is higher resolution than 16bit? Obviously it’s far lower resolution but with noise-shaped dither it’s perceptually about the same.

Why do you just say DSD uses 1 bit technique and don't mention the sampling frequency of DSD is 2.8224 MHz which is much much higher than the sampling frequency of CD which is only 44.1 KHz ?

 

[CiptoH]

In theory yes but in practice, not in the slightest. The only information below -120dB is noise and distortion.

The information removed when converting 24-bit to 16-bit starts at -96dB (16-bit) instead of -120dB (20-bit).
16-bit files can only accommodate a maximum of 16-bits and not 20-bits.

 

[gregorio]

Why do you just say DSD uses 1 bit technique and don't mention the sampling frequency of DSD is 2.8224 MHz which is much much higher than the sampling frequency of CD which is only 44.1 KHz ?

I say DSD uses 1bit because it does use 1bit and as you yourself quoted:

“If you are doing a linear stair-step digitization of an analog waveform, the number of bits used in the sampling determines the maximum dynamic range you can faithfully represent with the sampling.”

The DSD sampling frequency of 2.8mHz means we can encode audio frequency content to 1.4mHz but as human hearing only extends to 20kHz, all that extra frequency range is inaudible and there’s no recorded signal anywhere near 1.4mHz anyway. It is useful to have that very large inaudible frequency range in the case of 1bit however, because it provides an inaudible range to put the huge amount of noise that is generated by the noise-shaped dither process, which is what allows 1bit DSD to have a perceptual dynamic range of 120dB.

Rather than arguing with me and even with what you yourself have quoted, it would be better to ask if you don’t understand how digital audio works rather than argue. Also, it would be better to ask in the Sound Science subforum, where we’re allowed to directly quote the relevant science.

G

 

[gregorio]

16-bit files can only accommodate a maximum of 16-bits and not 20-bits.

Yes, and likewise 1bit DSD can only accommodate 1bit and therefore only 6dB of dynamic range. However, noise-shaped dither perceptually allows a dynamic range equivalent to 20bit, with both 1bit DSD and 16bit CD.

G

 

[CiptoH]

Also, humans can’t hear above 20kHz and adults typically lower than 16kHz. So it DOES mean all humans can’t hear the difference!

It's not just about the frequency response, it's about the difference in sound between 16-bit resolution recordings which are lower than 24-bit studio-quality recordings, which even though the difference in sound is not like day and night, is still there and 24-bit is still better.

 

[CiptoH]

The DSD sampling frequency of 2.8mHz means we can encode audio frequency content to 1.4mHz but as human hearing only extends to 20kHz, all that extra frequency range is inaudible and there’s no recorded signal anywhere near 1.4mHz anyway.

The basic principle of DSD is so different from PCM that the logic and calculations for PCM cannot be applied to DSD.

 

[CiptoH]

Yes, and likewise 1bit DSD can only accommodate 1bit and therefore only 6dB of dynamic range. However, noise-shaped dither perceptually allows a dynamic range equivalent to 20bit, with both 1bit DSD and 16bit CD.

Don't forget to mention that DSD has resolution around 24-bit.

 

[gregorio]

It's not just about the frequency response, it's about the difference in sound between 16-bit resolution recordings which are lower than 24-bit studio-quality recordings, which even though the difference in sound is not like day and night, is still there and 24-bit is still better.

The difference between 16/44 and say 24/96 is audio above 22kHz and noise/distortion below -120dB, both of which are inaudible to humans.

If you are hearing a difference between 16bit and 24bit versions, the only possible explanation is that either you’re just imagining a difference or you’re comparing different masters, where the 16bit version has had additional compression applied for example (which is not uncommon).

The basic principle of DSD is so different from PCM that the logic and calculations for PCM cannot be applied to DSD.

DSD is digital audio and therefore has to comply with the basic principles of digital audio the same as PCM does. The logic and calculations for PCM must therefore be applicable to DSD. In practice there is very little difference because PCM is initially sampled in DSD at very high sample rates using a few bits before being converted into PCM and most DSD recordings are converted to PCM for mixing and mastering.

Don't forget to mention that DSD has resolution around 24-bit.

But you should forget to mention that, because it’s false! You are arguing with me, the actual facts and even with what you yourself have quoted! Again, this is from the article YOU quoted:

“In terms of dynamic range, DSD is claimed to be around 120dB in the audible frequency range.”

24bit has a dynamic range of 144dB, DSD of around 120dB, NOT 144dB! 120dB is the dynamic range of 20bit, not 24bit.

G

 

[CiptoH]

DSD is digital audio and therefore has to comply with the basic principles of digital audio the same as PCM does. The logic and calculations for PCM must therefore be applicable to DSD.

DSD uses a very different type of modulation from PCM (sigma-delta modulation vs pulse code modulation.

"DSD (Direct Stream Digital) is an audiophile high-resolution audio format. It is a method of coding analog signal to digital form via sigma-delta modulation. Initially, it was designed an alternative to PCM (pulse code modulation), that used in CD-audio, WAV, FLAC, mp3 and other formats."
https://samplerateconverter.com/educational/what-dsd

DSD encoding

The pictures above is taken from :
https://samplerateconverter.com/educational/what-dsd

 

[gregorio]

DSD uses a very different type of modulation from PCM (sigma-delta modulation vs pulse code modulation.

Again, pretty much all professional PCM Analogue to Digital Converters, for around 30 years or so, initially digitise using a form of sigma/delta modulation and most DACs convert back to analogue using sigma/delta modulation, so how is that very different?

Plus, according to your own quote, DSD: “is a method of coding analog signal to digital form …”, So clearly DSD is digital audio and therefore must comply with the basic principles of digital audio (as must PCM of course).

Again, I don’t understand why you are arguing with the facts and even with what you yourself are quoting. It doesn’t make any rational sense. Clearly, you have a limited understanding of how digital audio works, so wouldn’t it be better to ask rather than argue?

G

 

[CiptoH]

But you should forget to mention that, because it’s false! You are arguing with me, the actual facts and even with what you yourself have quoted! Again, this is from the article YOU quoted:

“In terms of dynamic range, DSD is claimed to be around 120dB in the audible frequency range.”

24bit has a dynamic range of 144dB, DSD of around 120dB, NOT 144dB! 120dB is the dynamic range of 20bit, not 24bit.

You should quote in full so that the whole fact is clear :

"So how high resolution is a DSD recording? There’s no exact way of striking equivalence between PCM and DSD but it’s around the same as 24-bit/88.2kHz PCM".

"In terms of dynamic range, DSD is claimed to be around 120dB in the audible frequency range. In comparison, CD measures around 96dB, and 24-bit/192kHz recordings have a theoretical maximum of around 144dB".

https://www.whathifi.com/advice/what-dsd-audio-how-it-works-where-to-download-files-and-more

 

[Luke Skywalker]

The 16-bit 44.1KHz recording is just a downgraded version of the original 24-bit studio-master recording, in which there has been a decrease in resolution and dynamics range.
There are original 24-bit studio-master recordings available to consumers (especially for the most demanding audiophiles).

How about single bit DSDs? I have some that are 11.2 Mbps

Just saw you wrote that @CiptoH

Here is the real question. Does “burning in your headphones/IEMs” actually do anything? Most articles online say no, but I think most people here say yes

 

[CiptoH]

Again, pretty much all professional PCM Analogue to Digital Converters, for around 30 years or so, initially digitise using a form of sigma/delta modulation and most DACs convert back to analogue using sigma/delta modulation, so how is that very different?

Plus, according to your own quote, DSD: “is a method of coding analog signal to digital form …”, So clearly DSD is digital audio and therefore must comply with the basic principles of digital audio (as must PCM of course).

Again, I don’t understand why you are arguing with the facts and even with what you yourself are quoting. It doesn’t make any rational sense. Clearly, you have a limited understanding of how digital audio works, so wouldn’t it be better to ask rather than argue?

Not all calculations in PCM can be applied to DSD, for example, to determine the resolution in PCM is the same as the number of bits used, so if PCM uses 16 bit data, it means the resolution of the recording is 16 bit.
This can't be used for DSD because DSD uses 1 bit of sigma-delta modulation, if you use the formula for PCM above, the DSD recording resolution equals 1 bit, when in fact the DSD recording resolution is around 24 bits.

 

[gregorio]

You should quote in full so that the whole fact is clear :

The whole fact is clear: The dynamic range of DSD is around 120dB and the dynamic range of 24bit is 144dB. 120dB dynamic range and 144dB dynamic range are not the same, there’s 24dB difference, how is that not clear?

Also according to what you have quoted, bit depth defines the dynamic range, 8 bit = 48dB, 16bit = 96dB and 24bit = 144dB, so that’s 6dB dynamic range per bit. Therefore 1bit DSD has 6dB dynamic range and 120dB is equivalent to 20bit (excluding noise-shaped dither of course). I presume the article stated DSD is perceptually “around the same as 24/88” because 20bit was never a consumer format (only a professional format).

Again, you need to read what you have quoted and if you don’t fully understand it then ask!

G