"Passion for Sound" says they got meaningful differences between different USB cables- I want to try to reproduce but want feedback on approach

[xaxxon]

Video in question:

He takes a file, plays it through a dac via multiple USB cables and records it in Audacity. Then he inverts polarity and plays it against the source track. He gets noticeably different results between USB cables but multiple recordings using the same cable cancel out very well indicating that there isn't much in terms of run-to-run imperfections.

First, can anyone poke a huge hole in this other than "the data is faked"?

Assuming nothing obvious I would like to reproduce this test. He says he uses Focusright Scarlet 2i2 for recording - inexpensive and I have one on the way. I would be getting and using a variety of cables from what I have, buying some from amazon, maybe borrow some from local dealers, etc, and play and record the tracks through dacs I have and doing the same thing where I flip the polarity on the recordings and seeing if the difference between differences is significant.

I have a few dacs that I would be using including some that claim to have async modes to see if async matters. I would be using on-motherboard USB as a source and for the recording device.

Also, what kind of proof can I offer that the recordings are real? Just an uncut video showing the whole process for each cable/DAC tuplee?

​

Note: I don't think the guy is trustworthy at all, but he does make objective claims.

Dacs I own: dac3, x-sabre 3, d90se, dacmagic 200m. Maybe some others. Too bad I don't have @Amir 's infamously sensitive/broken schiit modi 2. :-\

 

[AudiOhm]

Don't waste your time and money...

Ohms

 

[DVDdoug]

Flipping the polarity and effectively subtracting can prove that there's no difference. But if you don't get silence it doesn't prove the sound is different and it's important to know that "the sound of the difference" is not the same as "the difference in the sound." The most obvious case is if you delay one copy by a few milliseconds before subtracting. There's no difference in the sound but if you subtract you'll get a "huge" difference file that's "louder" than the original. You'll also get "phasing" (comb filtering) and if you've done this kind of thing before you might know it's the result of a delay, but it doesn't sound like a delay.

If you do analog-to-digital conversion twice with the same-exact analog source you'll get different digital data. The main reason is that each sample is taken at one instant in time and it will be sampled at a different point along the waveform every time you digitize. The time/phase differences are "random". You can try to time-align the recordings after digitizing but it's quantized so once it's digitized you can't shift it by a fraction of a sample.

Plus there is always some analog noise and some time-drift. Even if you are very lucky and samples line-up perfectly at the beginning, the timing can be off by one or more (possibly many more) samples by the end of a song.

...So if you use the same USB cable twice the data won't null completely.

 

[AudiOhm]

Kimber Kable Review

Nordost Tyr 2 Review

Wireworld Starlight 7 Review

Ohms

 

[Rednaxela]

Have a look in the comments section.

 

[Blumlein 88]

Yes, don't waste your time. If you insist, the differences he found were due to tiny timing variations in the sample time of the Focusrite vs the playback of the track. Deltawave written by member pkane here will allow you to get proper results. It accounts for and adjust timing so you can get good deep null results. You'll find switching USB cables will leave you with mostly thermal noise with such comparisons.
Here is where to get the software. Free and excellent.

DeltaWave Audio Null Comparator | DeltaWave documentation

deltaw.org

 

[xaxxon]

Flipping the polarity and effectively subtracting can prove that there's no difference. But if you don't get silence it doesn't prove the sound is different and it's important to know that "the sound of the difference" is not the same as "the difference in the sound." The most obvious case is if you delay one copy by a few milliseconds before subtracting. There's no difference in the sound but if you subtract you'll get a "huge" difference file that's "louder" than the original. You'll also get "phasing" (comb filtering) and if you've done this kind of thing before you might know it's the result of a delay, but it doesn't sound like a delay.

If you do analog-to-digital conversion twice with the same-exact analog source you'll get different digital data. The main reason is that each sample is taken at one instant in time and it will be sampled at a different point along the waveform every time you digitize. The time/phase differences are "random". You can try to time-align the recordings after digitizing but it's quantized so once it's digitized you can't shift it by a fraction of a sample.

Plus there is always some analog noise and some time-drift. Even if you are very lucky and samples line-up perfectly at the beginning, the timing can be off by one or more (possibly many more) samples by the end of a song.

...So if you use the same USB cable twice the data won't null completely.

In the video the runs with the same cable nulled drastically more than the tests across cables. It was not a "perfect" or "not perfect" test it was very large difference in how well two runs with the same cable nulled.

 

[Chrispy]

Just the name Passion for Sound makes them sound like a waste of time.

 

[wwenze]

Common-mode noise in a computer is real btw. Even for a loop that goes out of one PCI slot and enters via another.

Is it real and measurable? Yes, from -100dBFS in the best PC to -70dB in the worst.

Is it circumventable? Also yes.

 

[Chrispy]

Common-mode noise in a computer is real btw. Even for a loop that goes out of one PCI slot and enters via another.

Is it real and measurable? Yes, from -100dBFS in the best PC to -70dB in the worst.

Is it circumventable? Also yes.

Or you could just not use a computer as a source, when I do it's wireless.

 

[raif71]

Just the name Passion for Sound makes them sound like a waste of time.

 

[Chrispy]

Uh huh.

 

[GXAlan]

@xaxxon
If you are going to measure cables, you will want to invest in a E1DA Cosmos since it is the cheapest premium ADC out there and sufficiently reliable to get good results.

Like everyone, I will tell you that measured differences does not translate into audible differences. That said, just as I may want to pay extra for an identical product that’s made in my hometown of Springfield over something made in Shelbyville for emotional not rational reasons, it’s understandable that one might be willing to pay extra for a better measuring product even if it’s not audible.

It’s hard to talk about good engineering of 123 vs 120 dB SINAD DACs and somehow say that it’s not equally clever that Wireworld was able to make a USB cable that actually improves measured performance (even though it doesn’t translate into audible performance).

Review and Measurements of Wireworld Starlight 7 USB Cable

This is a review and detailed measurements of Wireworld Starlight 7 flat USB cable. It is on kind loan from a member. The cost is USD $100 including Prime shipping for the 1 meter/40 inch length I am testing. For this review, I compared it to my current favorite, the 6 foot/1.5m version of...

www.audiosciencereview.com

There are two tools you can use to help, both from @pkane and both which are free.

The multitone test is good if you believe you can hear differences. If you think there is a difference in the bass, you can try generating multitones focused around that frequency spectrum. This is also good for making those 1 kHz test tones graphs that Amir used to show the difference with Wireworld cables or what I did here with even a mediocre ADC.

The DeltaWave tool lets you null differences between amps *and* run a PKmetric which helps you determine if something is audible or not. It will align samples for you as best as possible.

Both are found here:

DeltaWave Audio Null Comparator | DeltaWave documentation

deltaw.org

 

[fpitas]

Just the name Passion for Sound makes them sound like a waste of time.

I firmly grabbed my wallet.

 

[DAMUR]

Video in question:

He takes a file, plays it through a dac via multiple USB cables and records it in Audacity. Then he inverts polarity and plays it against the source track. He gets noticeably different results between USB cables but multiple recordings using the same cable cancel out very well indicating that there isn't much in terms of run-to-run imperfections.

First, can anyone poke a huge hole in this other than "the data is faked"?

Assuming nothing obvious I would like to reproduce this test. He says he uses Focusright Scarlet 2i2 for recording - inexpensive and I have one on the way. I would be getting and using a variety of cables from what I have, buying some from amazon, maybe borrow some from local dealers, etc, and play and record the tracks through dacs I have and doing the same thing where I flip the polarity on the recordings and seeing if the difference between differences is significant.

I have a few dacs that I would be using including some that claim to have async modes to see if async matters. I would be using on-motherboard USB as a source and for the recording device.

Also, what kind of proof can I offer that the recordings are real? Just an uncut video showing the whole process for each cable/DAC tuplee?

​

Note: I don't think the guy is trustworthy at all, but he does make objective claims.

Dacs I own: dac3, x-sabre 3, d90se, dacmagic 200m. Maybe some others. Too bad I don't have @Amir 's infamously sensitive/broken schiit modi 2. :-\

As a Network Administrator and ba in computer science plus an audiophile that has equipent in the 40-50k region solidstate and tube based plus paradigm high end monitors from 80s all redone by me and magnepan 3.7i speakers and 3.6r speakers plus more dacs then i can ever use including a 9038 and my main thing about cables is this: from a digital standpointthe only difference a usb cable will make is digital intererferance and with usb it has crc and parity data sent along with the digital audio signal. these are all ones and zeros that have to arrive at the other end in the exact same condition a 1 or a zero if there is interference it may turn a zero into a 1 but the cyclideric redundancy check (crc) and parity corrects these errors and makes it perfect everytime or it does not work simple. no cable will make one zero sound better then another 0 and u can run a usb cable in a same conduit as other cables or transmitters that can cause interference and even ac mains 120 volt less then 2mm away and that usb cable still sends that 1 and zero the same and on the other size recieves that one and zero. it may have to attempt the transmission multiple times if there are errors or high interference wich can create jitter or a noticeable delay but in the end it will correctly make it there the same. there is software that you can run to verify the transmission simular to how ip data works so it will show it had to resend the data due to errors but again if it cant make it to its destination correct it wont let it decode or you will here digital breakup which is percieved as silence or in mpeg compression digitization sounds and errors in block decoding making it look all jarbled. this was common back in the old sattelite days theses i am quite familiar with. so digital is only funked up by interference but any high end cable over 15-20 dollars will get your digital audio to its destination in one piece. i can actually show how this works by sending interference into a cable and having the sound output and showing the packet data on the send side and recieved side if everyone really wants to see this, but in short you can duplicate files purposely sending errors into it mid stream and the computer or chip will always correct it otherwise the data will be corrupted and unable to decode it. once it becomes analogue thats when you have to worry about interference so just use xlr and never worry again honestly! i wish digital was explained better to people because in the audio world people spend stupid amounts of money on snake oil things to make it sound better and it is nothing more then a scam for your money and the blacebo effect is very much real in humans unfortunately. i would be happy to explain more and please forgive spelling i have severe carpal tunnel and makes typing a bitch for me and cant be bother correcting the mistakes this kills me to write this but for people that dont understand digital data need to know a cable has to get it there perfect otherwise it doesent pass its certifications and unless u have tons of wireless transmissions besides your cables it wont be an issue for 99.999997 percent of people. Hope this helps.
Mike Muratagic
Network Administrator, MCP, CNA, A+,BA computer Science;

 

[Mat]

After doing a USB cable review video he's now arguing that USB is actually an analog signal, that the 'square waves' of digital information aren't ever perfect enough, and has unearthed 'studies' to objectively prove that $400 USB cables greatly improve the listening experience

re:

It's 100% an analog signal. Every digital audio transmission (other than optical) requires a fluctuating voltage. This is the best definition of an analog signal. The data is digital, but the transmission is not.

I've got some peer reviewed research papers about cable differences now so will be providing some data/measurement based analysis soon to explain why cables actually do make a difference. No opinion, no speculation, just data.

I guess too many people started calling him out on the USB thing so now he's making troll videos used to be a decent reviewer

 

[voodooless]

he's now arguing that USB is actually an analog signal

 

[Mat]

wow, didn't expect a full video on the topic

wow, didnt expect a fringe topic like this to have an entire video

"but he's completely bastardizing this into something that it's not"

The video summed things up quick.

 

[wwenze]

I want to add onto the "digital interpretation of the waveform" part: To the human eye, we can easily see that "if square is below line (or threshold) then interpret as zero, if square is above line then interpret as one". So people may think, "What about noise? the jigglies? what if there are short spikes within the squares that cut across the line even if most of the square is still above or below? What about ringing?"

Well here's the thing tho, we are identifying the square based on area. And guess what area actually is? Not width x height, that's a shortcut performed by humans due to limited processing capacity. Quote: A definite integral is the area under a curve between two fixed limits.

And it is easy to create an integrator circuit electronically. But we don't even have to - It works based on an RC network anyway. And RC is basically a low-pass filter. And it filters out the jigglies. And RC is everywhere. Which is why datasheets specify input hold times and stuff like that - Even if the incoming data isn't a perfect square, the input circuit, thanks to its own RC speed or bandwidth limitations, will need the incoming signal voltage to hold for a certain amount of time before that voltage is detected. In other words, the limitation of the system's bandwidth effectively already removes the jigglies and any ambiguity, for free. Likely with more consistency than human eyes.