• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Two reasonable DACs with a large PK Metric when tested with real music...

I don't see DSD as much better than MQA. 96/24 can do everything DSD can do and better without the downsides. I am not against music released in the format. Nor anyway we can get better masters. However just having DSD as an alternative complicates that for no real benefit. The discs that hold SACD can hold multi-channel 96/24 with much less data. I even understand DSD for a very brief period had some capability beyond redbook CD, but that period didn't last long before DVD audio discs were available and now with downloads and streams there is even less reason because it is so inefficient with the data. I don't see the need for more than 96 khz myself, but at least even DXD is a high rate PCM variant.
Please tell @pkane that you will still speak to him if he includes DSD512 in Multitone,you're his excuse that he doesn't!:(
 
1680034730520.jpeg


DeltaWave v2.0.8, 2023-03-28T13:08:56.6819271-07:00
Reference: D90-recording-trimmed.wav[R] 2568941 samples 48000Hz 24bits, stereo, MD5=00
Comparison: D90-recording-trimmed-plus1dB-repeat.wav[R] 2468304 samples 48000Hz 24bits, stereo, MD5=00
Settings:
Gain:True, Remove DC:True
Non-linear Gain EQ:False Non-linear Phase EQ: False
EQ FFT Size:65536, EQ Frequency Cut: 0Hz - 0Hz, EQ Threshold: -500dB
Correct Non-linearity: False
Correct Drift:True, Precision:30, Subsample Align:True
Non-Linear drift Correction:False
Upsample:False, Window:Kaiser
Spectrum Window:Kaiser, Spectrum Size:32768
Spectrogram Window:Hann, Spectrogram Size:4096, Spectrogram Steps:2048
Filter Type:FIR, window:Kaiser, taps:262144, minimum phase=False
Dither:False bits=0
Trim Silence:False
Enable Simple Waveform Measurement: False

Discarding Reference: Start=0.5s, End=5s
Discarding Comparison: Start=0.5s, End=5s

Initial peak values Reference: -11.658dB Comparison: -10.582dB
Initial RMS values Reference: -32.068dB Comparison: -32.29dB

Null Depth=26.489dB
X-Correlation offset: 163 samples
Drift computation quality, #1: Excellent (1.74μs)


Trimmed 0 samples ( 0.00ms) front, 0 samples ( 0.00ms end)


Final peak values Reference: -11.658dB Comparison: -12.022dB
Final RMS values Reference: -32.211dB Comparison: -33.73dB

Gain= 1.4401dB (1.1803x) DC=0 Phase offset=3.397088ms (163.06 samples)
Difference (rms) = -37.51dB [-56.2dBA]
Correlated Null Depth=78.97dB [35.92dBA]
Clock drift: -0.43 ppm


Files are NOT a bit-perfect match (match=0.07%) at 16 bits
Files are NOT a bit-perfect match (match=0%) at 24 bits
Files match @ 50.0006% when reduced to 6.15 bits


---- Phase difference (full bandwidth): 64.5315662089604°
0-10kHz: 58.27°
0-20kHz: 69.60°
0-24kHz: 64.53°
Timing error (rms jitter): 601.5μs
PK Metric (step=400ms, overlap=50%):
RMS=-50.4dBFS
Median=-51.9
Max=-43.4

99%: -44.03
75%: -49.57
50%: -51.95
25%: -53.92
1%: -57.4

gn=0.847217417445019, dc=0, dr=-4.29731023639261E-07, of=163.0602236678

DONE!

Signature: 3284cf201bc37a54e1cec592eaa9599e

RMS of the difference of spectra: -98.2786492403229dB
DF Metric (step=400ms, overlap=0%):
Median=-8.7dB
Max=-0.2dB Min=-20.5dB

1% > -19.8dB
10% > -14.56dB
25% > -11.03dB
50% > -8.71dB
75% > -6.58dB
90% > -3.93dB
99% > -0.15dB

Linearity 4.5bits @ 0.5dB error
DeltaWave v2.0.8, 2023-03-28T13:08:38.8167411-07:00
Reference: D90-recording-trimmed-plus1dB-repeat.wav[R] 2468304 samples 48000Hz 24bits, stereo, MD5=00
Comparison: SA10-recording-trimmed.wav[R] 2522627 samples 48000Hz 24bits, stereo, MD5=00
Settings:
Gain:True, Remove DC:True
Non-linear Gain EQ:False Non-linear Phase EQ: False
EQ FFT Size:65536, EQ Frequency Cut: 0Hz - 0Hz, EQ Threshold: -500dB
Correct Non-linearity: False
Correct Drift:True, Precision:30, Subsample Align:True
Non-Linear drift Correction:False
Upsample:False, Window:Kaiser
Spectrum Window:Kaiser, Spectrum Size:32768
Spectrogram Window:Hann, Spectrogram Size:4096, Spectrogram Steps:2048
Filter Type:FIR, window:Kaiser, taps:262144, minimum phase=False
Dither:False bits=0
Trim Silence:False
Enable Simple Waveform Measurement: False

Discarding Reference: Start=0.5s, End=5s
Discarding Comparison: Start=0.5s, End=5s

Initial peak values Reference: -10.582dB Comparison: -10.371dB
Initial RMS values Reference: -32.29dB Comparison: -31.389dB

Null Depth=27.242dB
Phase inverted
X-Correlation offset: 118 samples
Drift computation quality, #1: Excellent (0.39μs)


Trimmed 0 samples ( 0.00ms) front, 0 samples ( 0.00ms end)


Final peak values Reference: -10.582dB Comparison: -12.294dB
Final RMS values Reference: -32.29dB Comparison: -33.245dB

Gain= 1.923dB (1.2478x) DC=-0.00001 Phase offset=2.442266ms (117.229 samples)
Difference (rms) = -39.34dB [-51.65dBA]
Correlated Null Depth=26.06dB [29.46dBA]
Clock drift: 10.62 ppm


Files are NOT a bit-perfect match (match=0.06%) at 16 bits
Files are NOT a bit-perfect match (match=0%) at 24 bits
Files match @ 50.0093% when reduced to 6.09 bits


---- Phase difference (full bandwidth): 127.19493169824°
0-10kHz: 20.98°
0-20kHz: 114.20°
0-24kHz: 127.19°
Timing error (rms jitter): 119.5μs
PK Metric (step=400ms, overlap=50%):
RMS=-43.0dBFS
Median=-45.4
Max=-33.9

99%: -35.41
75%: -42.43
50%: -45.44
25%: -47.55
1%: -51.67

gn=0.801403551221381, dc=-1.15904125204398E-05, dr=1.06206201446066E-05, of=117.2287544101

DONE!

Signature: f977e5b4f7c8f8826cde26c3fdeb1ea8

RMS of the difference of spectra: -98.0238974473086dB
DF Metric (step=400ms, overlap=0%):
Median=-8.9dB
Max=-0.3dB Min=-20.6dB

1% > -19.3dB
10% > -15.69dB
25% > -11.6dB
50% > -8.93dB
75% > -6.44dB
90% > -4.46dB
99% > -0.2dB

Linearity 4.1bits @ 0.5dB error

Matching is inaccurate due to the inherent noise of the UMIK-1 and the low volume recording.

You can see that there are portions where it matches but the fit is poor.
1680035245344.png




and it grossly matches
1680035344069.png



And here is the PK Metric

1680035777341.png


There still suggests that there is greater difference between the two source units than there is between repeated measures of the identical unit.

The real question is what the minimum clinically significant difference in PK Metric is....
 
View attachment 275583

DeltaWave v2.0.8, 2023-03-28T13:08:56.6819271-07:00
Reference: D90-recording-trimmed.wav[R] 2568941 samples 48000Hz 24bits, stereo, MD5=00
Comparison: D90-recording-trimmed-plus1dB-repeat.wav[R] 2468304 samples 48000Hz 24bits, stereo, MD5=00
Settings:
Gain:True, Remove DC:True
Non-linear Gain EQ:False Non-linear Phase EQ: False
EQ FFT Size:65536, EQ Frequency Cut: 0Hz - 0Hz, EQ Threshold: -500dB
Correct Non-linearity: False
Correct Drift:True, Precision:30, Subsample Align:True
Non-Linear drift Correction:False
Upsample:False, Window:Kaiser
Spectrum Window:Kaiser, Spectrum Size:32768
Spectrogram Window:Hann, Spectrogram Size:4096, Spectrogram Steps:2048
Filter Type:FIR, window:Kaiser, taps:262144, minimum phase=False
Dither:False bits=0
Trim Silence:False
Enable Simple Waveform Measurement: False

Discarding Reference: Start=0.5s, End=5s
Discarding Comparison: Start=0.5s, End=5s

Initial peak values Reference: -11.658dB Comparison: -10.582dB
Initial RMS values Reference: -32.068dB Comparison: -32.29dB

Null Depth=26.489dB
X-Correlation offset: 163 samples
Drift computation quality, #1: Excellent (1.74μs)


Trimmed 0 samples ( 0.00ms) front, 0 samples ( 0.00ms end)


Final peak values Reference: -11.658dB Comparison: -12.022dB
Final RMS values Reference: -32.211dB Comparison: -33.73dB

Gain= 1.4401dB (1.1803x) DC=0 Phase offset=3.397088ms (163.06 samples)
Difference (rms) = -37.51dB [-56.2dBA]
Correlated Null Depth=78.97dB [35.92dBA]
Clock drift: -0.43 ppm


Files are NOT a bit-perfect match (match=0.07%) at 16 bits
Files are NOT a bit-perfect match (match=0%) at 24 bits
Files match @ 50.0006% when reduced to 6.15 bits


---- Phase difference (full bandwidth): 64.5315662089604°
0-10kHz: 58.27°
0-20kHz: 69.60°
0-24kHz: 64.53°
Timing error (rms jitter): 601.5μs
PK Metric (step=400ms, overlap=50%):
RMS=-50.4dBFS
Median=-51.9
Max=-43.4

99%: -44.03
75%: -49.57
50%: -51.95
25%: -53.92
1%: -57.4

gn=0.847217417445019, dc=0, dr=-4.29731023639261E-07, of=163.0602236678

DONE!

Signature: 3284cf201bc37a54e1cec592eaa9599e

RMS of the difference of spectra: -98.2786492403229dB
DF Metric (step=400ms, overlap=0%):
Median=-8.7dB
Max=-0.2dB Min=-20.5dB

1% > -19.8dB
10% > -14.56dB
25% > -11.03dB
50% > -8.71dB
75% > -6.58dB
90% > -3.93dB
99% > -0.15dB

Linearity 4.5bits @ 0.5dB error
DeltaWave v2.0.8, 2023-03-28T13:08:38.8167411-07:00
Reference: D90-recording-trimmed-plus1dB-repeat.wav[R] 2468304 samples 48000Hz 24bits, stereo, MD5=00
Comparison: SA10-recording-trimmed.wav[R] 2522627 samples 48000Hz 24bits, stereo, MD5=00
Settings:
Gain:True, Remove DC:True
Non-linear Gain EQ:False Non-linear Phase EQ: False
EQ FFT Size:65536, EQ Frequency Cut: 0Hz - 0Hz, EQ Threshold: -500dB
Correct Non-linearity: False
Correct Drift:True, Precision:30, Subsample Align:True
Non-Linear drift Correction:False
Upsample:False, Window:Kaiser
Spectrum Window:Kaiser, Spectrum Size:32768
Spectrogram Window:Hann, Spectrogram Size:4096, Spectrogram Steps:2048
Filter Type:FIR, window:Kaiser, taps:262144, minimum phase=False
Dither:False bits=0
Trim Silence:False
Enable Simple Waveform Measurement: False

Discarding Reference: Start=0.5s, End=5s
Discarding Comparison: Start=0.5s, End=5s

Initial peak values Reference: -10.582dB Comparison: -10.371dB
Initial RMS values Reference: -32.29dB Comparison: -31.389dB

Null Depth=27.242dB
Phase inverted
X-Correlation offset: 118 samples
Drift computation quality, #1: Excellent (0.39μs)


Trimmed 0 samples ( 0.00ms) front, 0 samples ( 0.00ms end)


Final peak values Reference: -10.582dB Comparison: -12.294dB
Final RMS values Reference: -32.29dB Comparison: -33.245dB

Gain= 1.923dB (1.2478x) DC=-0.00001 Phase offset=2.442266ms (117.229 samples)
Difference (rms) = -39.34dB [-51.65dBA]
Correlated Null Depth=26.06dB [29.46dBA]
Clock drift: 10.62 ppm


Files are NOT a bit-perfect match (match=0.06%) at 16 bits
Files are NOT a bit-perfect match (match=0%) at 24 bits
Files match @ 50.0093% when reduced to 6.09 bits


---- Phase difference (full bandwidth): 127.19493169824°
0-10kHz: 20.98°
0-20kHz: 114.20°
0-24kHz: 127.19°
Timing error (rms jitter): 119.5μs
PK Metric (step=400ms, overlap=50%):
RMS=-43.0dBFS
Median=-45.4
Max=-33.9

99%: -35.41
75%: -42.43
50%: -45.44
25%: -47.55
1%: -51.67

gn=0.801403551221381, dc=-1.15904125204398E-05, dr=1.06206201446066E-05, of=117.2287544101

DONE!

Signature: f977e5b4f7c8f8826cde26c3fdeb1ea8

RMS of the difference of spectra: -98.0238974473086dB
DF Metric (step=400ms, overlap=0%):
Median=-8.9dB
Max=-0.3dB Min=-20.6dB

1% > -19.3dB
10% > -15.69dB
25% > -11.6dB
50% > -8.93dB
75% > -6.44dB
90% > -4.46dB
99% > -0.2dB

Linearity 4.1bits @ 0.5dB error

Matching is inaccurate due to the inherent noise of the UMIK-1 and the low volume recording.

You can see that there are portions where it matches but the fit is poor.
View attachment 275588



and it grossly matches
View attachment 275590


And here is the PK Metric

View attachment 275591

There still suggests that there is greater difference between the two source units than there is between repeated measures of the identical unit.

The real question is what the minimum clinically significant difference in PK Metric is....
Your Difference RMS was 37-39 db for all of these even same vs same. Roughly the level I achieve with the same approach using completely different speakers than yours.
 
Here’s what I wonder though… for the actually DSD recorded and mastered music, maybe the the INABILITY to throw dynamic range compressors and all sorts of fancy processing onto the mix makes the resulting sound better. Many photographers get better results on analog film compared to digital *because* they are more thoughtful and deliberate in their photo making when each exposure costs money.

The assumption is that you have a skilled recording engineer — but an unskilled recording engineer may be protected against bad decisions when working in DSD.

Hah! If there were many (more than a handful) recording engineers that used DSD for recording directly from a mic with no processing, I might think that there could be some benefit here. Except that most of the DSD content is mastered as PCM (even when recorded as DSD) and then converted back to DSD/SACD for distribution, so the advantage is dubious at best.
 
...what is DSD...???...am I better off not knowing...?!?...

geicocavemen2.jpg
 
View attachment 275583

DeltaWave v2.0.8, 2023-03-28T13:08:56.6819271-07:00
Reference: D90-recording-trimmed.wav[R] 2568941 samples 48000Hz 24bits, stereo, MD5=00
Comparison: D90-recording-trimmed-plus1dB-repeat.wav[R] 2468304 samples 48000Hz 24bits, stereo, MD5=00
Settings:
Gain:True, Remove DC:True
Non-linear Gain EQ:False Non-linear Phase EQ: False
EQ FFT Size:65536, EQ Frequency Cut: 0Hz - 0Hz, EQ Threshold: -500dB
Correct Non-linearity: False
Correct Drift:True, Precision:30, Subsample Align:True
Non-Linear drift Correction:False
Upsample:False, Window:Kaiser
Spectrum Window:Kaiser, Spectrum Size:32768
Spectrogram Window:Hann, Spectrogram Size:4096, Spectrogram Steps:2048
Filter Type:FIR, window:Kaiser, taps:262144, minimum phase=False
Dither:False bits=0
Trim Silence:False
Enable Simple Waveform Measurement: False

Discarding Reference: Start=0.5s, End=5s
Discarding Comparison: Start=0.5s, End=5s

Initial peak values Reference: -11.658dB Comparison: -10.582dB
Initial RMS values Reference: -32.068dB Comparison: -32.29dB

Null Depth=26.489dB
X-Correlation offset: 163 samples
Drift computation quality, #1: Excellent (1.74μs)


Trimmed 0 samples ( 0.00ms) front, 0 samples ( 0.00ms end)


Final peak values Reference: -11.658dB Comparison: -12.022dB
Final RMS values Reference: -32.211dB Comparison: -33.73dB

Gain= 1.4401dB (1.1803x) DC=0 Phase offset=3.397088ms (163.06 samples)
Difference (rms) = -37.51dB [-56.2dBA]
Correlated Null Depth=78.97dB [35.92dBA]
Clock drift: -0.43 ppm


Files are NOT a bit-perfect match (match=0.07%) at 16 bits
Files are NOT a bit-perfect match (match=0%) at 24 bits
Files match @ 50.0006% when reduced to 6.15 bits


---- Phase difference (full bandwidth): 64.5315662089604°
0-10kHz: 58.27°
0-20kHz: 69.60°
0-24kHz: 64.53°
Timing error (rms jitter): 601.5μs
PK Metric (step=400ms, overlap=50%):
RMS=-50.4dBFS
Median=-51.9
Max=-43.4

99%: -44.03
75%: -49.57
50%: -51.95
25%: -53.92
1%: -57.4

gn=0.847217417445019, dc=0, dr=-4.29731023639261E-07, of=163.0602236678

DONE!

Signature: 3284cf201bc37a54e1cec592eaa9599e

RMS of the difference of spectra: -98.2786492403229dB
DF Metric (step=400ms, overlap=0%):
Median=-8.7dB
Max=-0.2dB Min=-20.5dB

1% > -19.8dB
10% > -14.56dB
25% > -11.03dB
50% > -8.71dB
75% > -6.58dB
90% > -3.93dB
99% > -0.15dB

Linearity 4.5bits @ 0.5dB error
DeltaWave v2.0.8, 2023-03-28T13:08:38.8167411-07:00
Reference: D90-recording-trimmed-plus1dB-repeat.wav[R] 2468304 samples 48000Hz 24bits, stereo, MD5=00
Comparison: SA10-recording-trimmed.wav[R] 2522627 samples 48000Hz 24bits, stereo, MD5=00
Settings:
Gain:True, Remove DC:True
Non-linear Gain EQ:False Non-linear Phase EQ: False
EQ FFT Size:65536, EQ Frequency Cut: 0Hz - 0Hz, EQ Threshold: -500dB
Correct Non-linearity: False
Correct Drift:True, Precision:30, Subsample Align:True
Non-Linear drift Correction:False
Upsample:False, Window:Kaiser
Spectrum Window:Kaiser, Spectrum Size:32768
Spectrogram Window:Hann, Spectrogram Size:4096, Spectrogram Steps:2048
Filter Type:FIR, window:Kaiser, taps:262144, minimum phase=False
Dither:False bits=0
Trim Silence:False
Enable Simple Waveform Measurement: False

Discarding Reference: Start=0.5s, End=5s
Discarding Comparison: Start=0.5s, End=5s

Initial peak values Reference: -10.582dB Comparison: -10.371dB
Initial RMS values Reference: -32.29dB Comparison: -31.389dB

Null Depth=27.242dB
Phase inverted
X-Correlation offset: 118 samples
Drift computation quality, #1: Excellent (0.39μs)


Trimmed 0 samples ( 0.00ms) front, 0 samples ( 0.00ms end)


Final peak values Reference: -10.582dB Comparison: -12.294dB
Final RMS values Reference: -32.29dB Comparison: -33.245dB

Gain= 1.923dB (1.2478x) DC=-0.00001 Phase offset=2.442266ms (117.229 samples)
Difference (rms) = -39.34dB [-51.65dBA]
Correlated Null Depth=26.06dB [29.46dBA]
Clock drift: 10.62 ppm


Files are NOT a bit-perfect match (match=0.06%) at 16 bits
Files are NOT a bit-perfect match (match=0%) at 24 bits
Files match @ 50.0093% when reduced to 6.09 bits


---- Phase difference (full bandwidth): 127.19493169824°
0-10kHz: 20.98°
0-20kHz: 114.20°
0-24kHz: 127.19°
Timing error (rms jitter): 119.5μs
PK Metric (step=400ms, overlap=50%):
RMS=-43.0dBFS
Median=-45.4
Max=-33.9

99%: -35.41
75%: -42.43
50%: -45.44
25%: -47.55
1%: -51.67

gn=0.801403551221381, dc=-1.15904125204398E-05, dr=1.06206201446066E-05, of=117.2287544101

DONE!

Signature: f977e5b4f7c8f8826cde26c3fdeb1ea8

RMS of the difference of spectra: -98.0238974473086dB
DF Metric (step=400ms, overlap=0%):
Median=-8.9dB
Max=-0.3dB Min=-20.6dB

1% > -19.3dB
10% > -15.69dB
25% > -11.6dB
50% > -8.93dB
75% > -6.44dB
90% > -4.46dB
99% > -0.2dB

Linearity 4.1bits @ 0.5dB error

Matching is inaccurate due to the inherent noise of the UMIK-1 and the low volume recording.

You can see that there are portions where it matches but the fit is poor.
View attachment 275588



and it grossly matches
View attachment 275590


And here is the PK Metric

View attachment 275591

There still suggests that there is greater difference between the two source units than there is between repeated measures of the identical unit.

The real question is what the minimum clinically significant difference in PK Metric is....

I suspect that reflections and resonances are a larger reason for the low-ish null numbers.
 
Hah! If there were many (more than a handful) recording engineers that used DSD for recording directly from a mic with no processing, I might think that there could be some benefit here. Except that most of the DSD content is mastered as PCM (even when recorded as DSD) and then converted back to DSD/SACD for distribution, so the advantage is dubious at best.
I only know of him,he is somehow dedicated to this.
(I never heard his music,I know nothing)

 
I suspect that reflections and resonances are a larger reason for the low-ish null numbers.
Plus it’s raining today :)
Better to ignore. DSD is like MQA, only less lossy and just as inconvenient ;)
As the owner of the SA-10, my recommendation is as follows.

1) Do you have an existing SACD library?
No —> stop. You are done. Do not buy a SACD player.
Yes —> is your current SACD player broken? If not, do not buy another SACD player.

2) Do you enjoy physical media and “unplugging” over a local DLNA server?
No —> stop. You are done. Buy a used Sony Blu-Ray player and extract your SACD collection to a network attached storage
Yes —> you might want a SACD player

3) Are you OK listening to your physical SACDs on your home theater system?
Yes —> stop. You are done. Get a Sony BD or UHD disc player which will transcode DSD to PCM over HDMI with “Sony approved” algorithms
No —> you might want a SACD player.

4) Do you like vinyl LP playback?
Yes —> you should probably get a SACD player because you enjoy anachronisms and experience more than pure sound quality.
No —> You should sell your SACD collection. You won’t miss it.
 
Last edited:
On Qobuz, when I see 192k recordings, they sometimes have high levels of ultrasonic noise, typically in the 40k to 90k frequency range. When I read the liner notes, they were inevitably recorded in DSD. The conversion to PCM for streaming seems to have insufficient filtering.
 
On Qobuz, when I see 192k recordings, they sometimes have high levels of ultrasonic noise, typically in the 40k to 90k frequency range. When I read the liner notes, they were inevitably recorded in DSD. The conversion to PCM for streaming seems to have insufficient filtering.
They probably are just transcoding it without extra filtering. One could even say if you like DSD this is the right way. I wouldn't say that.
 
Better to ignore. DSD is like MQA, only less lossy and just as inconvenient ;)

...I have become very accomplished at ignorance...I have been ignoring MQA since I first heard of it...
 
Back
Top Bottom