Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[pschelbert]

For me -very important - I have various DSD versions of albums that I also have in 24/192 flac and there are differences. In general (and it could be the way my audio is set up) I find the DSD just tends to flow beter.

yes, the difference is the mastering. companies like Sony, Universal, Warner think they can improve manipulating, compressing (make it louder) and so on. DSD is differently mastered from the same source when it sounds different.

Prove yourself:
http://www.2l.no/hires/index.html

[pschelbert]

here from an example of how DSD is done:

"Provenance: Recorded to DSD64 and mixed through analog console to DSD64 (Tr.5,6,7,9,10,12) and DSD128 (Tr.1,2,3,4,8,11,13,14). The DSD64 mixes have been upsampled in the Double DSD version for your convenience."

Yeah, the converison is here done via analog. Two conversions, not so conving regarding tecnical quality.

[michael123]

here from an example of how DSD is done:

"Provenance: Recorded to DSD64 and mixed through analog console to DSD64 (Tr.5,6,7,9,10,12) and DSD128 (Tr.1,2,3,4,8,11,13,14). The DSD64 mixes have been upsampled in the Double DSD version for your convenience."

Yeah, the converison is here done via analog. Two conversions, not so conving regarding tecnical quality.

Same is done many times for PCM, where some tracks are not available for example as 96/24, but as 44.1/24. You don't want to switch sample rate in the middle of the album

[CLW6]

I bought Media Center specifically for the DSD support.

[pschelbert]

Same is done many times for PCM, where some tracks are not available for example as 96/24, but as 44.1/24. You don't want to switch sample rate in the middle of the album

once digital it must stay digital. No reason to go to analog.

[pschelbert]

Same is done many times for PCM, where some tracks are not available for example as 96/24, but as 44.1/24. You don't want to switch sample rate in the middle of the album

The bright guys do upsampling and sellling it as 96/24, making more money

[Fitzcaraldo215]

here from an example of how DSD is done:

"Provenance: Recorded to DSD64 and mixed through analog console to DSD64 (Tr.5,6,7,9,10,12) and DSD128 (Tr.1,2,3,4,8,11,13,14). The DSD64 mixes have been upsampled in the Double DSD version for your convenience."

Yeah, the converison is here done via analog. Two conversions, not so conving regarding tecnical quality.

I think you need a deeper understanding of recording practices.  That is not at all typical practice for the classical albums I buy.  I would not buy them if they did.  Yes, for editing, short sippers of DSD are sometimes converted to DXD, then back again after the edits for final release on DSD.

With pop or other genres, who knows?  But, those other genres are not typically recorded in DSD, because that is not as easily processed or manipulated, as is the norm in those genres.  Cookie Marenco is a notable exception to that.

With all due respect to those who find something special in DSD128, 256 or higher, I find precious little to rave about.  DSP room EQ, applied to PCM conversions of DSD, are much more sonically significant in my view. 

And, upsampling from the native recorded sampling rate, whether in DSD or PCM, does not hold much water for me.  It is a turn off.  You cannot add back something that was never captured in the first place. 

[pschelbert]

I absolutely agree.
The way it should be done.

[Axilian]

One thing to take note of is the current design of most DAC chips - they are actually based on processing delta sigma signals. (and mainly have been since the  first Phillips 'bitstream' CD players back in the 90's) (It's certainly the way ESS 90XX DAC chips work)

So in fact a PCM signal actually requires additional processing compared to DSD (a delta sigma format)  as the PCM signal has to go through a stage of interpolation into a delta-sigma format before conversion

As for upsampling there's really no point as you cannot create actual resolution by upsampling - if  you have say 4 Hz and say upsample to 8 Hz really the only real data is the 4 data points the rest is really only guesswork. This was one of the problems with the first upsamplers where the algorithm was so basic that the upsampling actual cause 'smearing' of the original sound. Of course things have  moved on but the basic  fact that you cannot create real data remains

The best way to describe it is say you have 8 points in music of which points 1.3.5.7 are real sample points. If you have a sharp transient signal at point 6 which fades by point 7 - upsampling will never see it as it will interpolate from point 5-7 as a smooth transition a(and hence still look like the 4 bit signal. The only way to catch the transient to sample fast enough to actual catch it as a real data point 

The same theory applies to any sort of digital measuring device (I deal with sampling instrumentation and calibration of such on a daily basis)

[RD James]

I think you need a deeper understanding of recording practices.  That is not at all typical practice for the classical albums I buy.  I would not buy them if they did.  Yes, for editing, short sippers of DSD are sometimes converted to DXD, then back again after the edits for final release on DSD.

To clarify this statement, for anyone that doesn't know: "DXD" is 24-bit 352.8 kHz PCM.
Once it has been converted to PCM for editing it should never be re-encoded to DSD, since DSD encoding is a lossy process.
If you are editing in PCM, you should be recording in PCM to begin with.
 
There is nothing superior about a high sample rate 1-bit format compared to modern multi-bit PCM.
An argument could maybe be made in favor of DSD when compared against CD-quality 16-bit 44.1kHz PCM audio, but not anything newer than that.

One thing to take note of is the current design of most DAC chips - they are actually based on processing delta sigma signals. (and mainly have been since the  first Phillips 'bitstream' CD players back in the 90's) (It's certainly the way ESS 90XX DAC chips work)
So in fact a PCM signal actually requires additional processing compared to DSD (a delta sigma format)  as the PCM signal has to go through a stage of interpolation into a delta-sigma format before conversion

Yes, but these are multi-bit converters, not 1-bit converters (technically they combine a lot of 1-bit converters to create a multi-bit device) so there is no benefit to feeding them a 1-bit audio format.

As for upsampling there's really no point as you cannot create actual resolution by upsampling - if  you have say 4 Hz and say upsample to 8 Hz really the only real data is the 4 data points the rest is really only guesswork. This was one of the problems with the first upsamplers where the algorithm was so basic that the upsampling actual cause 'smearing' of the original sound. Of course things have  moved on but the basic  fact that you cannot create real data remains

Upsampling can improve things, such as DSP performed on the signal, however.
For example: you can use far more aggressive noise shaping when dithering if you are upsampling to a very high sample rate, than playback at 44.1 kHz.
Rather than having a lot of noise at ~21kHz, I can move it out to say 84kHz if I upsample to 176.4 kHz.

[Axilian]

Hadn't thought about the noise band shifting that is a reason to upsampling (in fact that one of the reason DSD 128 was introduced - to shift the noise band, DSD64 noise band was considered to be a little close to the audio range.

[Fitzcaraldo215]

To clarify this statement, for anyone that doesn't know: "DXD" is 24-bit 352.8 kHz PCM.
Once it has been converted to PCM for editing it should never be re-encoded to DSD, since DSD encoding is a lossy process.
If you are editing in PCM, you should be recording in PCM to begin with.
 
There is nothing superior about a high sample rate 1-bit format compared to modern multi-bit PCM.
An argument could maybe be made in favor of DSD when compared against CD-quality 16-bit 44.1kHz PCM audio, but not anything newer than that.
Yes, but these are multi-bit converters, not 1-bit converters (technically they combine a lot of 1-bit converters to create a multi-bit device) so there is no benefit to feeding them a 1-bit audio format.
Upsampling can improve things, such as DSP performed on the signal, however.
For example: you can use far more aggressive noise shaping when dithering if you are upsampling to a very high sample rate, than playback at 44.1 kHz.
Rather than having a lot of noise at ~21kHz, I can move it out to say 84kHz if I upsample to 176.4 kHz.

I hate to tell you this, but converting snippets of a DSD recording to DXD for edits, then back again to DSD for release as DSD is and has been done routinely.  I have this info from well qualified recording engineers.  Not, to worry though, they are very short clips, and they do it only as necessary.  If you can hear any of it on DSD recordings, though, let the label know.  But, I doubt it is audible or stands out in any way.  And, it is a pragmatic necessity given the limitations of DSD DAWs.

Also, for whatever reason, some engineering teams just prefer the sound of DSD, whatever you or the rest of us may think and in spite of the difficulties in editing and mastering in DSD.  They stake their brand identity and reputation on it.  Others, of course, disagree.

I personally remain agnostic in this "religious war".  I prefer using DSP in playback and view that as much more important than DSD vs. PCM, even if it involves conversion of DSD to PCM.

I

[RD James]

I hate to tell you this, but converting snippets of a DSD recording to DXD for edits, then back again to DSD for release as DSD is and has been done routinely.  I have this info from well qualified recording engineers.  Not, to worry though, they are very short clips, and they do it only as necessary.  If you can hear any of it on DSD recordings, though, let the label know.  But, I doubt it is audible or stands out in any way.  And, it is a pragmatic necessity given the limitations of DSD DAWs.

Well that's kind of my point.
There are really two types of DSD content:

• DSD which has gone through at least one PCM conversion stage in its editing.
• DSD where the production has been done with analog tools to avoid PCM.

You don't really have a "100% DSD workflow" because it's not a format that can be edited natively.
Either of those scenarios would have been better served by a PCM workflow since its multi-bit nature does allow for editing.

Also, for whatever reason, some engineering teams just prefer the sound of DSD, whatever you or the rest of us may think and in spite of the difficulties in editing and mastering in DSD.  They stake their brand identity and reputation on it.  Others, of course, disagree.

I can't really argue against people's preferences, but from an objective point of view there are distortions with DSD that cannot be eliminated as a result of it being a 1-bit format.
You can increase the sample rate enough that they will effectively be inaudible, but cannot be eliminated.

I personally remain agnostic in this "religious war".  I prefer using DSP in playback and view that as much more important than DSD vs. PCM, even if it involves conversion of DSD to PCM.

I agree with you that DSP is important in a modern playback system - which is why I believe that DSD should be left in the past, since it has to be converted to PCM for that anyway.
DSD has just been used as a marketing point to sell something unique to audiophiles, regardless of its technical properties. Your DAC doesn't do DSD playback? Better buy our shiny new DAC that does!
Same thing with the lies about MQA being better than PCM when it's really just a lossy compression scheme for PCM.

[pschelbert]

seems that the DSD and MQA marketing hype is outside Europe as at the HighEnd2018 in Munich I have not seen much (last year, 2017) it was a lot more.
May be the music-industry will be forced to learn.

For practical reasons it good to narrow the number of formats. I even digitize vinyl to flac if the music is something special (no Michael Jackson I do not digitize )

Conversion tools are appreciated though, especially I use:
DSD to PCM
ALAC to flac
AAC to mp3

Its all in MC so no need to change anything.

[tyler69]

Unless there are other requriements on your side for MP3, I think AAC might be the better option: https://www.winxdvd.com/resource/aac-vs-mp3.htm

[pschelbert]

My requirement is standard to the max. Shall play on every device.
Technically AAC is better but proprietory.

[RD James]

AAC is an open proprietary standard, just like MP3 - though the patents for MP3 did expire last year.
If you want open, you'd be using Vorbis/Opus - though they don't have the wide support of AAC or MP3.
MP3 will have the broadest compatibility, but there are very few devices that won't play an AAC file these days.
Compression/Quality is much higher with AAC than MP3. High bitrate AAC borders on audibly lossless at less than half the size of FLAC.

[pschelbert]

I did standardize on mp3 long before AAC was open. Too late too bad for apple.