For Those Naysayers...

So what type of new music today isn't extensively edited? Classical and rock sure are. Folk? So the question is still how many "DSD" recordings are true DSD and not converted back and forth for editing?

It would seem to me that the only real DSD recordings are--and coming back to Peter's post--were originally analog.

Yes, those that were either balanced and mixed in analog, prior to conversion to DSD, and/or those post processed on a Sonoma meet that criteria.

Myles, you first need to define your term Editing. If you mean all post processing and sweetening, then you're correct, except for Sonoma DAW produced. If you mean splices, then any recording not be-balanced in the digital domain is a pure native DSD recording, save the bunch of 50 millisecond splices that were processed in PCM

Most DSD recorded classical are not sweetened, but are certainly spliced (edited). All rock is both highly post processed, and edited.
 
That is the description of how (Sony) Sonoma operates; conversion of a 1-bit two level Pulse Density Modulation data STREAM (DSD) to an 8-bit PDM data STREAM. It is not PCM, which is a 2's compliment weighted, stand alone value, WORD based encoding system. An X-bit wide PDM data STREAM is simply one of a practically infinite number of encoding schemes derivable from a Delta-Sigma Modulator(s), and its number and type of feedback loops within. The key point however is that 8-bit two level "DSD Wide" is both mathematically operatable (editable), and because it operates at the same sample rate as the original DSD, there's there's no decimation filtering required as in lower sampling rate PCM conversion.

Agree. DSD-wide is a hybrid format that does not involve a sample rate conversion.

Pyramix uses an entirely different scheme. They use the Philips developed alternative to the Sony (Oxford) developed DSD Wide editing format, using 352.8KHz PCM, named DXD. As Bruce points out though, for simple edits, only the actual edit crossfade interval is rendered using DXD, leaving the remaining DSD file untouched. If a level has to be changed, let alone ANY post processing sweetening, then the entire file becomes DXD, if preformed in Pyramix.

And DXD is basically 24/352.8 PCM, right?
 
And DXD is basically 24/352.8 PCM, right?

Exactly, except that it's specified as 32 bits, with 64-bit internal math processing. Its advantage is that it opened to Pyramix users a wide variety of industry standard PCM post processing plug-ins that could be/were adapted to operate at 8X PCM sampling rate.
 
Exactly, except that it's specified as 32 bits, with 64-bit internal math processing. Its advantage is that it opened to Pyramix users a wide variety of industry standard PCM post processing plug-ins that could be/were adapted to operate at 8X PCM sampling rate.

How is DSD converted to DXD?
 
How is DSD converted to DXD?

Bruce is far more qualified to answer this than I. Within Pyramix there are conversion routines with different filters, in addition to popular apps like Weiss Saracon. Also, at least one A/D converter, the DAD AX24, which Merging incorporates as their Sphynx A/D Converter, does the conversion from a Delta-Sigma Modulator front end internally to DXD. This is the converter 2L uses for their recordings.
 
Exactly, except that it's specified as 32 bits, with 64-bit internal math processing.

Ah, thanks. All the descriptions I have come across have described it as 24 bit, 352.8kHz.

So, in summary, the only editing operations that can be done on native 1-bit DSD are time-based (as in cutting and splicing in the days of tape), but anything that actually requires a waveform transformation (change in volume/gain, equalizing, filtering etc.) requires a conversion to either hi-res PCM (DXD), short-word hybrid (DSD-wide) or analog and back.
 
Yes, because the individual 1-bit DSD samples convey no amplitude information in themselves, just change information based on the previous sample. It's a derivative of the actual analog information, and therefore must be integrated to bring it to a factor that can be operated mathematically. That's basically what converting to PCM accomplishes.

In of itself, conversion to PCM is of no consequence to the data, as long as there's enough bits in the computation to eliminate round-off errors. It's the decimation filtering required in lowering the sampling rates that does the sound quality damage. That's one of the reasons converting to DSD Wide (8-bit two level PDM) at the same sampling rate sounds better with music containing real ambiance information (which the ear/brain is very sensitive due to our life's experience hearing things naturally).

For that same reason, it's why 352.8KHz PCM sounds better than 192/176KHz, which sounds better than 88.2/96KHz etc. The faster the sampling rate, all other things being equal, the less invasive the decimation filtering.
 
In of itself, conversion to PCM is of no consequence to the data, as long as there's enough bits in the computation to eliminate round-off errors.

While I agree with you, I am not sure that everybody in the "DSD sounds better than PCM" camp agrees with us.

For that same reason, it's why 352.8KHz PCM sounds better than 192/176KHz, which sounds better than 88.2/96KHz etc. The faster the sampling rate, all other things being equal, the less invasive the decimation filtering.

Here it might again be good to add a "in my opinion" or "sounds better *to me*", as I don't think there is enough objective evidence that the improvement from sampling rates is audible once you get to rates where modern decimation filtering algorithms can do their work in a non-intrusive manner. But that is a different discussion, and not related to DSD.
 
Here it might again be good to add a "in my opinion" or "sounds better *to me*", as I don't think there is enough objective evidence that the improvement from sampling rates is audible once you get to rates where modern decimation filtering algorithms can do their work in a non-intrusive manner.

You of course, are correct about my not limiting the statement to one of my experience. From my experience though, once you edit in these formats all day long, you develop an "ear" for how they individually sound, especially on electrostatic speakers with natural acoustic music. You can pick them out quickly, and you develop a subconscious grading scheme. YMMV however.
 
Last edited:
While I agree with you, I am not sure that everybody in the "DSD sounds better than PCM" camp agrees with us.

Who's "us"? I believe you either cherry-picked my statement of "no consequence of DSD conversion to PCM", or misunderstood. It's the following sentence about decimation filters that's relevant. There's most certainly a consequence in converting DSD at any supported sampling rate to any lower supported PCM sampling rate. I'm definitely in the "DSD sounds better than PCM" camp.
 
Last edited:
(...) From my experience though, once you edit in these formats all day long, you develop an "ear" for how they individually sound, especially on electrostatic speakers with natural acoustic music. You can pick them out quickly, and you develop a subconscious grading scheme. YMMV however.

Do you own SoundLabs?
 
Some clarification.

@DSD_Audio: Sony HAP-Z1ES hi-rez media player will upsample any format to double-DSD as well as handling native single-DSD and double-DSD files.
 
It's a derivative of the actual analog information, and therefore must be integrated to bring it to a factor that can be operated mathematically.

Alright!
I knew my high school calculus would have practical applications ;)
 
Who's "us"? I believe you either cherry-picked my statement of "no consequence of DSD conversion to PCM", or misunderstood. It's the following sentence about decimation filters that's relevant. There's most certainly a consequence in converting DSD at any supported sampling rate to any lower supported PCM sampling rate. I'm definitely in the "DSD sounds better than PCM" camp.

My apologies if I misunderstood, but the way I read what you are saying is that it is downsampling that can be harmful to sound quality, not the conversion to PCM as such. So a conversion to PCM with a high enough sample rate to not require a downsampling is OK (as in DXD), but a downsampling to a lower sample frequency, independent of format, can have consequences to sound quality. If that is not what you are saying, I'd appreciate being set right.
 
My apologies if I misunderstood, but the way I read what you are saying is that it is downsampling that can be harmful to sound quality, not the conversion to PCM as such. So a conversion to PCM with a high enough sample rate to not require a downsampling is OK (as in DXD), but a downsampling to a lower sample frequency, independent of format, can have consequences to sound quality. If that is not what you are saying, I'd appreciate being set right.

And what about the number of conversions?
 
And what about the number of conversions?

I guess we can agree on a general rule that conversions are best avoided, despite some of them not having much audible effect.
 
...but the way I read what you are saying is that it is downsampling that can be harmful to sound quality, not the conversion to PCM as such. So a conversion to PCM with a high enough sample rate to not require a downsampling is OK (as in DXD), but a downsampling to a lower sample frequency, independent of format, can have consequences to sound quality.

We agree, except that DXD is also subject to the effects of decimation, but less so than lower sampling rate PCM. It's a matter of degree.

The conversion from DSD to PCM has two parts; The algorithm to change the format, and the decimation filtering to preclude imaging back those frequencies in the DSD data stream above the Nyquist frequency of the PCM sampling rate.

PCM is like a strip of film, with individual stand alone frame pictures sequenced together at some frame per second rate (sampling rate). DSD (1-bit two level) is a serialized stream, where only a change/no change indicator from the previous sample time level is present. DSD and PCM can be transcoded, and given sufficient processing bit depth to eliminate any round-off errors, can be processed back and forth indefinitely, with no loss.

The problem is that presently there are no facilities to play, or make useful the above transcoded PCM at the DSD sampling rate. It must be processed into a new lower sampling rate, which includes decimation filtering. The Nyquist frequency of any format is half the sampling rate. 64fs DSD is 1.41MHz, and DXD is 176.4KHz. That's an 8X reduction, meaning all frequency energy above 176.4KHz must be filtered out (at least -120dB). That's a pretty steep filter if the desire is to have an audio bandpass to at least 40KHz, let alone 100KHz. And of course, it gets worse for lower sampling rate PCM.

Can we hear this, and does it make any difference? What a great hobby and business!
 
That's an 8X reduction, meaning all frequency energy above 176.4KHz must be filtered out (at least -120dB). That's a pretty steep filter if the desire is to have an audio bandpass to at least 40KHz, let alone 100KHz. And of course, it gets worse for lower sampling rate PCM.

I agree with the basic premise, and the filtering is pretty demanding because of the high amounts of HF noise energy caused by the noise shaping of DSD. But especially because of the noise shaping I don't see any reason not to filter at a lower frequency, say 22 kHz, and that allows for 3 octaves between 22 and 176.4 kHz - a fairly benign filter slope, compared to the brickwall filtering required for 44.1 kHz redbook/CD.

Can we hear this, and does it make any difference?

Indeed, that is the essential question.

What a great hobby and business!

Great hobby, and mostly a great business, as long as the business isn't too dominated by snake oil.
 

About us

  • What’s Best Forum is THE forum for high end audio, product reviews, advice and sharing experiences on the best of everything else. This is THE place where audiophiles and audio companies discuss vintage, contemporary and new audio products, music servers, music streamers, computer audio, digital-to-analog converters, turntables, phono stages, cartridges, reel-to-reel tape machines, speakers, headphones and tube and solid-state amplification. Founded in 2010 What’s Best Forum invites intelligent and courteous people of all interests and backgrounds to describe and discuss the best of everything. From beginners to life-long hobbyists to industry professionals, we enjoy learning about new things and meeting new people, and participating in spirited debates.

Quick Navigation

User Menu

Steve Williams
Site Founder | Site Owner | Administrator
Ron Resnick
Site Owner | Administrator
Julian (The Fixer)
Website Build | Marketing Managersing