CD player vs. DAC sound - can you hear a difference? (listening test)

[b4nt]

I answered no. Sounds/files seem similar.

I would have to concentrate, to check if little noises I hear are exact the same.

If I point approx where I noticed some (present and audible in both files), Audacity tells me the tracks are different

 

[pma]

To me, Adobe Audition is still the best tool to analyze the files and differences, if they are properly matched. I does not introduce unpredictable math manipulation.

I do not agree that the spectra of the files start to differ from about 5kHz. I am stating that it starts from about 20kHz. This is the only noticeable difference between the files, a result of different frequency responses of the filters of the CD player and the DAC, not speaking about a constant shift in microseconds (which introduces apparent phase shift) which is the result of non-simultaneous non-synchronized sampling of the CD output and DAC output.

I have prepared a file in Audition where the left channel contains Alpha left channel and the right channel contains Beta left channel. Time analysis and spectrum analysis near sample no. 2110540 was done, where the difference between the files Alpha and Beta is the biggest.



Second, my intent is to show that there would be for almost sure no audible difference between properly designed and working digital signal sources (like here we use 18 years old CD player) and that the race for isolated parameters is useless. If there is a sound difference, it may be due to secondary technical issues like signal ground loops, intermodulations with ultra sound noise components and EMI induced issues. This is not covered by tests of individual components under clinical conditions. So if someone states he can hear a difference between 2 SOTA DACs, he should look for mistakes in his setup first.

 

[pma]

I answered no. Sounds/files seem similar.

I would have to concentrate, to check if little noises I hear are exact the same.

If I point approx where I noticed some (present and audible in both files), Audacity tells me the tracks are different


View attachment 140805

You cannot compare sample values only as they are not taken at the same instant. It is not a loopback test. CD player plays freely and is recorded by the ADC with non-synchronized clock. Then the DAC plays and is again recorded by the ADC with non-synchronized clock. As far as we fulfill the Nyquist criterion - the original signal can be reconstructed from its samples if the sampling frequency is >2x higher than the highest frequency of the sampled signal. Which is fulfilled here. But don't forget the reconstruction is made by reconstruction filter and you have nothing like this in Audacity. To compare sample values only is grossly misleading.

 

[b4nt]

Second, my intent is to show that there would be for almost sure no audible difference between properly designed and working digital signal sources (like here we use 18 years old CD player) and that the race for isolated parameters is useless. If there is a sound difference, it may be due to secondary technical issues like signal ground loops, intermodulations with ultra sound noise components and EMI induced issues. This is not covered by tests of individual components under clinical conditions. So if someone states he can hear a difference between 2 SOTA DACs, he should look for mistakes in his setup first.

Other part tracks seem the same, according to Audacity. Why are tracks different where I heard noises (in both tracks)?

 

[b4nt]

But don't forget the reconstruction is made by reconstruction filter and you have nothing like this in Audacity. To compare sample values only is grossly misleading.

Ok, I'll listen better there, to check. Noises being in treebles or so. But I answered files seem similar during playback.

Edit: wider than trebles, seems to be low level string noises.

 

[b4nt]

You cannot compare sample values only as they are not taken at the same instant.

CD is at 44.1/16, files are at 48/24. Could be interesting to record again, but at 96k?

 

[pkane]

So how are you doing sub-sample alignment? Does it interpolate between samples?


That will be great.

Yes, subsample alignment to better than 1/1000 of a sample. There’s no interpolation or upsampling, the offset is applied by modifying phase.

New version released with spectrogram log-frequency support.

 

[pkane]

To me, Adobe Audition is still the best tool to analyze the files and differences, if they are properly matched. I does not introduce unpredictable math manipulation.

I do not agree that the spectra of the files start to differ from about 5kHz. I am stating that it starts from about 20kHz. This is the only noticeable difference between the files, a result of different frequency responses of the filters of the CD player and the DAC, not speaking about a constant shift in microseconds (which introduces apparent phase shift) which is the result of non-simultaneous non-synchronized sampling of the CD output and DAC output.

I have prepared a file in Audition where the left channel contains Alpha left channel and the right channel contains Beta left channel. Time analysis and spectrum analysis near sample no. 2110540 was done, where the difference between the files Alpha and Beta is the biggest.

View attachment 140811 View attachment 140812

Second, my intent is to show that there would be for almost sure no audible difference between properly designed and working digital signal sources (like here we use 18 years old CD player) and that the race for isolated parameters is useless. If there is a sound difference, it may be due to secondary technical issues like signal ground loops, intermodulations with ultra sound noise components and EMI induced issues. This is not covered by tests of individual components under clinical conditions. So if someone states he can hear a difference between 2 SOTA DACs, he should look for mistakes in his setup first.

The spectra do start to diverge around 5kHz, but with very tiny, inaudible differences. Phase does start to deviate around 5kHz:

Here's the spectra differences for the whole track:

Zoomed in above 8kHz:

Difference in tracks spectra:

Spectrogram using an exaggerated palette meant to show minor differences:

Using a normal (Rainbow) palette:

 

[AnalogSteph]

The spectra do start to diverge around 5kHz, but with very tiny, inaudible differences. Phase does start to deviate around 5kHz:

The phase going south exponentially up there looks suspiciously like a time delay, i.e. imperfect time alignment. That said, it could also be the result of an IIR lowpass... does one of the two devices have a Cirrus DAC?

 

[pkane]

The phase going south exponentially up there looks suspiciously like a time delay, i.e. imperfect time alignment. That said, it could also be the result of an IIR lowpass... does one of the two devices have a Cirrus DAC?

No, time delay was removed. This is the non-linear part.

 

[b4nt]

To compare sample values only is grossly misleading.

I do not think those samples (minimal amount and variations) would produce the same output signal. I will consider this small difference is due to any noise that poped in the recording equipments. And assume I noticed those variations because I focused on the low level sounds/noises I heard around: my mistake.

Files sounded the same, didn't check much further, neither if I see other such differences inside else tracks parts. Might just be additionnal noise, EMI or other.

 

[pma]

Just tried Deltawave by @pkane . Looks like a useful tool, congrats. I will have to spend some time to learn all capabilities of the program.

 

[pma]

May I humbly ask those who are posting "Yes" answer in the poll to attach the ABX result as well? Or at least to describe the sound difference and sample file time when it was audible. Just to make sure it was not only a yes click . Thank you.

 

[cany89]

@pma since this poll is closed, can you share details?

 

[pma]

@pma since this poll is closed, can you share details?

Sure, it was the direct analog output from Vincent CD-S3 player (Alpha) vs. Vincent CD-S3==>SPDIF coax==>DacMagicPlus analog output (Beta).

 

[cany89]

Sure, it was the direct analog output from Vincent CD-S3 player (Alpha) vs. Vincent CD-S3==>SPDIF coax==>DacMagicPlus analog output (Beta).

Good gear for sure! I listened to a bit. Since the poll was closed I haven't spent enough time but the first impression is I heard no difference tbh.