Home Psychoacoustics Blind Tests: starting with DACs of the Apple USB-C dongle vs Chord Mojo

[Panasonyum]

Background:
I was unsatisfied with a couple of previous blind tests of DAC dongles vs bigger DACs. For example, Archimag's test used recordings of DACs as opposed to participants listening to actual DACs https://archimago.blogspot.com/2024/05/high-end-dac-blind-listening-results.html, other tests used DACs+amps and headphones (e.g., "Epic Multiple Blind Tests! Can They Tell The Differences?" on youtube). I wanted to listen to the actual sources and to eliminate amps.

Setup:
Apple USB-C Dongle connected to a PC; and PC optical out to Chord Mojo 1. Both sources are connected to an RCA switcher which is wired to my active studio monitor speakers Adam A5X. https://i.postimg.cc/vmc1nhj6/mojo.jpg

Procedure:
My test was to be done in the following way. There are N trials, each trial consisting of the same song section played twice. The experimenter chooses if it is played from different sources (e.g., dongle then Mojo) or on the same source (eg., dongle twice). Volume is set to zero in the beginning of each song, the participant is free to adjust it. The participant ranks the first song as 10, second song as 11, 10 or 9 =better, same, worse. The experimenter adds up all the scores for both sources, divides by the number of trials and we have a result.

Unfortunately I could not convince my partner that the needs of science were greater than hers, so I did the "blind test" myself - thus omitting the hard "any trial could contain two instances of the same source" case.

Yesterday, I've attached all the audio cables to an RCA switch box, and this morning I realized that I could not remember which switch position corresponded to which source - hence my decision to run the test. Apart from the physical button, I've used a script to switch between the optical out to Mojo and the Apple dongle sources in Foobar. When I clicked it, it switched between the two sources without it telling me which of them was active (there was merely no sound if it did not correspond to the RCA switcher position).

Spoiler: Foobar Script:

Install the Spider Monkey Panel component. To add this scripting ability, you must first install the foo_spider_monkey_panel component. You can download it from the foobar2000 components page and install it by going to File > Preferences > Components > Install.
Locate your device names. In foobar2000, go to File > Playback > Device. Note the exact names of the output devices you want to switch between. They will be listed under the DS (DirectSound), WASAPI, or ASIO sub-menus, depending on your setup and installed components. For example: DS: Speakers (Realtek(R) Audio).
Scripting with the Spider Monkey Panel
Add a new Spider Monkey Panel: Right-click on your foobar2000 layout and select Add new panel > Spider Monkey Panel.
Open the panel configuration: Right-click the new panel and select Configure panel.
Insert the script: In the script editor window, enter the following code. This example shows a simple script to cycle between two devices, triggered by a click.

// Configuration
var devices = [
"Playback/Device/Realtek Digital Output (Realtek USB Audio) [exclusive]",
"Playback/Device/Headphones (USB-C to 3.5mm Headphone Jack Adapter) [exclusive]",
//"Playback/Device/Primary Sound Driver",
];
var currentDeviceIndex = 0;

// A label to display and click
function on_paint(gr) {
gr.FillSolidRect(0, 0, this.width, this.height, 0); // Black background
gr.DrawString(
"Switch Device",
gdi.Font("Arial", 14),
0xFFFFFFFF,
10,
10,
this.width,
this.height
);
}

// A function to handle the click and change the device
function on_mouse_lbtn_up(x, y) {
currentDeviceIndex = (currentDeviceIndex + 1) % devices.length;
var deviceName = devices[currentDeviceIndex];
fb.RunMainMenuCommand(deviceName);

// Optional: Show a popup to confirm the device change
//fb.ShowPopupMessage("Switched to: " + deviceName.split("/").pop(), "Device Switch");
}

Every song started with the volume dial on the speakers set to 0 or near 0. I've adjusted loudness for every song to where it felt good - this was different for every song and every trial. Volume settings were very close and I could not guess the source from fumbling with the volume dial.

Results:
Out of 30 trials, Mojo won on 18 trials. 10 trials were a draw. Apple won 2 trials.

• Chi-square statistic: 12.8
• p-value: 0.00166
• The p-value (0.00166) is much smaller than the typical significance level of 0.05.
• This means I reject the null hypothesis and conclude that there is a statistically significant difference in the preference between Mojo and Apple Dongle.

Here are the scores for 30 trials (flat = flat RCA switch position, this turned out to correspond to the Apple dongle; Sticking = the switch was sticking out, this turned out to be Mojo): https://i.postimg.cc/XY65mqLy/Mojo-experiment.jpg

Discussion:
It was hard to find differences on all tracks, there were no "night and day" tracks. Acoustical instruments, vocals and acoustic drums sound "more flowing" on Mojo. Electronic music sounded extremely similar, although maybe I was still getting accustomed to the test.

Shortcomings:
1).I always knew that each trial had 2 different sources, this made things easier - this is a biggie, and I still hope to convince my partner to donate 30 minutes of her life to redo this.
2).Although the volume dial settings were very similar for both sources (about the 1st mark on the drawn scale- depending on the song), and I did not look at this scale when setting volume, I may have developed a subconscious feel for which source corresponded to which volume setting. Asking my partner to go louder or quieter would also mitigate this. I don't want to rely on precise volume matching as changing volume for each song and genre is a natural way to listen to music and I tend to doubt the precision of those measurements.

 

[welwynnick]

Congratulations, I think this is what Audio Science should be about - getting away from your keyboard and doing controlled listening tests, rather than parroting "all DACs sound the same" in response to every discussion about audibility. The former is very difficult and is to be congratulated, and the latter is easy and is not.
Having said that I do have some comments and questions:

For example, Archimag's test used recordings of DACs as opposed to participants listening to actual DACs https://archimago.blogspot.com/2024/05/high-end-dac-blind-listening-results.html, other tests used DACs + amps and headphones (e.g., "Epic Multiple Blind Tests! Can They Tell The Differences?" on youtube). I wanted to listen to the actual sources and to eliminate amps.

I agree about recording DACs. I always found that adding a ADDA loop in a system made you listen to the loop instead of the source.

If you're using Adam active speakers, you're not eliminating the amps, because your speakers have amps (and they might not be the greatest).

What was your music source, in particular, what was the digital audio format - standard or high resolution?

I agree about the material. Simple, well-recorded, acoustic arrangements are a better test than electronic music or movie soundtracks.

If you want to convince the scientific community that this is a meaningful test, then you must have the levels matched accurately.

(If you use a different volume setting with each test then you lose control of the test. Some people will insist that you need to match levels to within +/-0.1 or 0.2 dB, though I don't subscribe to that particular obsession. I used to strive for objective assessment long before Audio Science Review started, and I tried doing sensitivity assessments by doing AB comparisons with different products with level differences of +2dB, +1dB, 0dB, -1dB and -2dB. None of that ever made any difference to my subjective preference, it just made one source sound quieter or louder, that's all. No-matter, the keyboard warriors round here who consider themselves scientific (yes I am being cynical about that) will insist on level matching for every test.)

Congratulations again, Nick

 

[Panasonyum]

Thanks for the kind words

If you're using Adam active speakers, you're not eliminating the amps, because your speakers have amps (and they might not be the greatest).

I should have phrased it better, I meant i was eliminating the uncontrolled variable of amps. In my case, the same amp was used for both DACs. (And I love my Adams blindly )

What was your music source, in particular, what was the digital audio format - standard or high resolution?

A collection of 44100Hz FLAC files and a couple of 320kbs MP3s with female vocals, Mojo won there as well.

If you want to convince the scientific community that this is a meaningful test, then you must have the levels matched accurately.

I wrote about this in the Shortcomings section. Since each test started with zero volume, I could have preferred a source that required, e.g., bigger turn of the volume dial. So I could have focused on the rate of dialing as opposed to the actual sound quality. Imo it was unlikely, but it is a shortcoming. Loudness matching would require a microphone of unquestionable precision, otherwise the results could also be dismissed - "your mic wasn't accurate enough, therefore one of the sources played louder".

I think this is what Audio Science should be about - getting away from your keyboard and doing controlled listening tests

Yep! And I would emphasize listening tests, not measuring tests. As honest to god blind tests as one has time for, where people reliably prefer sound of one audio device over another. Not based on their graphs, but on their music/sound reproduction. And you're right of course about the "all DACs sound the same" narrative, but there is also "my super special DAC is super special" narrative as well

 

[jmdesignz2]

Now which of these measures best?

What are the voltage outputs and their SNR/THD? Is there a correlation to what OP preferred?

 

[Panasonyum]

Now which of these measures best?

What are the voltage outputs and their SNR/THD? Is there a correlation to what OP preferred?

Let's assume that the blind test preferences and the SNR/THD measurements do match in one case. But does this mean they will always match? What should we do if there is reliable preference for a source with inferior SNR? Or if there is no reliable preference despite objective differences in SNR? To me the answer is obvious: I want the source that I perceive as the best sounding and I do not want to pay for the difference that i cannot perceive.

I am certainly not advocating abandoning physics/acoustics for creating and testing audio equipment - that is simply impossible. But having psychoacoustic tests to calibrate our sanity is also not amiss, JNDs (just noticeable differences) have been invented for a reason

PS Not really adding or retracting from the points above, I doubt that, e.g., a Steinway grand piano sound can be captured in just a few easy to read graphs. To what extent this also applies to speakers, amps, and DACs is an open question, but I doubt the similarity is none.

 

[RandomEar]

An interesting experiment! It's still slightly flawed due to not being double blind and having different levels, as you mentioned yourself. But that should be easy to fix.

Comparing the devices and the setup, there's a couple of midlly interesting differences:

	Apple Dongle	Chord Mojo
SINAD	99 dB	102 dB
Frequency response (20 Hz - 20 kHz)	+0.02/-0.12 dB [1]	+0.01/-0.20 [2]
Reconstruction filter	Minimum phase [1]	Linear phase [2]
Jitter	Approx. -110 dB	Approx. -130 dB
Linearity (+-0.5 dB)	-107 dB	-115 dB
Crosstalk	-84 dB [1]	-82 dB [2]
Power (300 Ω / 33 Ω)	3.6 mW / 31 mW	71 mW / 410 mW
Output impedance	0.9 Ω	0.7 Ω (manufacturer claims 0.075 Ω)
Peak output voltage	1.0 V	4.9 V [2]

The input impedance of the A5X is listed as 30 kΩ, so that should be fine for pretty much any source. Overall, the dongle and the Mojo have pretty similar specs. There's nothing standing out concerning noise or distortion. One clear difference is the ability of the Mojo to deliver nearly 5 V of output. If you accidentally ran the comparison with the Apple dongle maxing out its 1 V limit causing occasional clipping, this could be a plausible explanation for audible differences. If you repeat the experiment, I would recommend to check and equalize the output level of both DACs.

The setup with PC->USB dongle is also susceptible to noise and ground loops. USB power on PCs is notorious for noise problems. So at least checking for any audible noise with the dongle route active at normal listening levels would be advisable.

 

[solawind]

didn't those A5X speakers have their own ADC - DAC conversion inside? Or are they pure analog speakers
and what about sampling rate for devices in windows settings

 

[oleg87]

An interesting experiment! It's still slightly flawed due to not being double blind and having different levels, as you mentioned yourself. But that should be easy to fix.

Comparing the devices and the setup, there's a couple of midlly interesting differences:

	Apple Dongle	Chord Mojo
SINAD	99 dB	102 dB
Frequency response (20 Hz - 20 kHz)	+0.02/-0.12 dB [1]	+0.01/-0.20 [2]
Reconstruction filter	Minimum phase [1]	Linear phase [2]
Jitter	Approx. -110 dB	Approx. -130 dB
Linearity (+-0.5 dB)	-107 dB	-115 dB
Crosstalk	-84 dB [1]	-82 dB [2]
Power (300 Ω / 33 Ω)	3.6 mW / 31 mW	71 mW / 410 mW
Output impedance	0.9 Ω	0.7 Ω (manufacturer claims 0.075 Ω)
Peak output voltage	1.0 V	4.9 V [2]

The input impedance of the A5X is listed as 30 kΩ, so that should be fine for pretty much any source. Overall, the dongle and the Mojo have pretty similar specs. There's nothing standing out concerning noise or distortion. One clear difference is the ability of the Mojo to deliver nearly 5 V of output. If you accidentally ran the comparison with the Apple dongle maxing out its 1 V limit causing occasional clipping, this could be a plausible explanation for audible differences. If you repeat the experiment, I would recommend to check and equalize the output level of both DACs.

The setup with PC->USB dongle is also susceptible to noise and ground loops. USB power on PCs is notorious for noise problems. So at least checking for any audible noise with the dongle route active at normal listening levels would be advisable.

The Apple dongle has 3dB of headroom beyond 1Vrms, I'm not set up to measure its performance at the limit, but it shouldn't clip.

If I'm understanding the setup correctly, OP's "blinding" relied on not remembering which switch position is which, and I'm not sure if any attempt to match device output volume was made, as one source requiring a slightly different position for a subjectively "appropriate" volume would be a considerable tell. This seems like a pretty serious experimental flaw as it effectively makes it "pseudonymous" rather than blind. Better than sighted I reckon, but properly controlled tests when you're dealing with hardware rather than digital recordings are tricky.

 

[solawind]

If you repeat the experiment, I would recommend to check and equalize the output level of both DACs.

Yes, that would be a more correct way.

 

[voodooless]

It's still slightly flawed due to not being double blind and having different levels, as you mentioned yourself.

It’s not slightly flawed, it’s utterly flawed. Level matching is of utmost importance in blind tests!

Loudness matching would require a microphone of unquestionable precision, otherwise the results could also be dismissed - "your mic wasn't accurate enough, therefore one of the sources played louder".

A microphone is no good. Measure the voltage with a steady tone with a DMM at the amp or speaker terminal and match that to less than .1V.

 

[Panasonyum]

Some good constructive criticism there! Here is some acceptance and some pushback from my side

The setup with PC->USB dongle is also susceptible to noise and ground loops. USB power on PCs is notorious for noise problems. So at least checking for any audible noise with the dongle route active at normal listening levels would be advisable.

The sound was clean even at very low volumes for both.

and what about sampling rate for devices in windows settings

Both used Foobar's exclusive mode - as seen in the script

A microphone is no good. Measure the voltage is a steady tone with a DDM in the amp or speaker terminal and match that to less than .1V.

This is an excellent tip, thank you! Mojo's lineout operates at 3V I now read. Reducing its gain would activate its headphone amp as they share the same 3.5mm socket. But, I suppose, I could adjust volume in Foobar to get the required voltage and add this adjustment to my hidden source switching script above.

It’s not slightly flawed, it’s utterly flawed. Level matching is of utmost importance in blind tests!

If one source played at higher volume through all the tests, then the experiment would be utterly flawed. This was not the case here, there was no systematic bias. Each source came both first and second on different trials and the source that followed it had a chance to compensate by "asking" for higher volume. (x60 for all songs in the experiment, so I could have played at window shaking volume by the end )

as one source requiring a slightly different position for a subjectively "appropriate" volume would be a considerable tell.

The flat or the sticking out position of the RCA switcher button was the ultimate tell I knew every single time that I was switching to a different source, I just did not know which one. I similarly did not know which volume dial position corresponded to which source. Moreover, volume dial position was harder to judge because:

1). I did not return the volume dial to its absolute zero for every song to confuse my muscle memory. Different songs required different volume which confused my muscle memory more.
2) The dial travel from zero to listening volume was about 5mm, the zero volume has play of another 2mm. These small adjustments with their small differences were difficult to differentiate by touch. Both sources played at my comfortable level at around 5mm mark.

Basically, I would still defend my method as good enough for cases where one cannot use microphone or multimeter. Asking another person to go "louder please! louder, louder, stop!" would further decouple the participant. But, even without this extra step, if it is not known which exact source corresponds to which physical volume dial setting, then all you learn is that you're playing a different source - as was the case here.

 

[DanielT]

That Apple USB-C via Android is hopeless. Minimal output power. I basically had to crank up a 30 watt amp to max to get to almost normal volume.
The sound quality Apple USB-C via Android vs. another DAC + Android was perhaps audible because the small 30 watt amplifier operated under different conditions. Nothing I explored further because I gave the Apple USB-C to my friend who has Apple stuff.
What I mean is that Apple USB-C as such is most likely completely transparent with Apple as a source/OS. Otherwise, well...

Note that my testing is all on Windows. I attempted to test on my Samsung S8+ and got odd results. Using both Google dongles, all they did was route the not so good sound of the internal DAC through the dongle (???). With Apple dongle I got music but level was very low. So if you plan to use the Apple dongle on an Android phone, you should do some compatibility testing.

Review: Apple vs Google USB-C Headphone Adapters

This is a review, detailed measurements and comparison of Apple's USB-C adapter to the current and last version of Google Pixel headphone adapters. The Apple adapter costs just $9 including one day shipping for free. The Google dongle costs $12. Not that any of these are large by any stretch...

www.audiosciencereview.com

Yes I saw that you @Panasonyum used PC just thought I'd mention this with Android as general information. By the way, thanks for taking the time to put together a test.

 

[Multicore]

If you accidentally ran the comparison with the Apple dongle maxing out its 1 V limit causing occasional clipping,

At what dBFS does the Apple dongle clip?

 

[solawind]

Basically, I would still defend my method as good enough for cases where one cannot use microphone or multimeter.

It would be very interesting for many, i believe, if you could redo this test with proper volume matching at the AC ports level. Not going to argue that your method of adjusting volume is flawed, but it is another variable in the test and could somehow (volume control non linearities etc) affect its results

 

[RandomEar]

At what dBFS does the Apple dongle clip?

Thinking about it again, it should probably never clip if designed correctly - it will simply not output anymore than the maximum voltage of roughly 1 Vpeak (if I'm reading Amir's review correctly) when playing a 0 dB digital signal. At least as long as it is feeding a high impedance input like on the A5X.

 

[RandomEar]

Some good constructive criticism there! Here is some acceptance and some pushback from my side

The sound was clean even at very low volumes for both.

You need high volumes to test this, because your amp in the A5X is amplifying the noise from your source. Therefore, at low volume it's less likely that any noise is audible at all (apart from the self-noise of the A5X amp, which will be identical for all sources).

If one source played at higher volume through all the tests, then the experiment would be utterly flawed. This was not the case here, there was no systematic bias. Each source came both first and second on different trials and the source that followed it had a chance to compensate by "asking" for higher volume. (x60 for all songs in the experiment, so I could have played at window shaking volume by the end )

There could absolutely be systemic bias, because the dongle starts out at a much lower volume than the Mojo, meaning your volume tuning is influenced by the anchoring effect. As pointed out by other members, level matching is really, really important.

 

[dominikz]

Background:
I was unsatisfied with a couple of previous blind tests of DAC dongles vs bigger DACs. For example, Archimag's test used recordings of DACs as opposed to participants listening to actual DACs https://archimago.blogspot.com/2024/05/high-end-dac-blind-listening-results.html, other tests used DACs+amps and headphones (e.g., "Epic Multiple Blind Tests! Can They Tell The Differences?" on youtube). I wanted to listen to the actual sources and to eliminate amps.

Setup:
Apple USB-C Dongle connected to a PC; and PC optical out to Chord Mojo 1. Both sources are connected to an RCA switcher which is wired to my active studio monitor speakers Adam A5X. https://i.postimg.cc/vmc1nhj6/mojo.jpg

Procedure:
My test was to be done in the following way. There are N trials, each trial consisting of the same song section played twice. The experimenter chooses if it is played from different sources (e.g., dongle then Mojo) or on the same source (eg., dongle twice). Volume is set to zero in the beginning of each song, the participant is free to adjust it. The participant ranks the first song as 10, second song as 11, 10 or 9 =better, same, worse. The experimenter adds up all the scores for both sources, divides by the number of trials and we have a result.

Unfortunately I could not convince my partner that the needs of science were greater than hers, so I did the "blind test" myself - thus omitting the hard "any trial could contain two instances of the same source" case.

Yesterday, I've attached all the audio cables to an RCA switch box, and this morning I realized that I could not remember which switch position corresponded to which source - hence my decision to run the test. Apart from the physical button, I've used a script to switch between the optical out to Mojo and the Apple dongle sources in Foobar. When I clicked it, it switched between the two sources without it telling me which of them was active (there was merely no sound if it did not correspond to the RCA switcher position).

Spoiler: Foobar Script:

Install the Spider Monkey Panel component. To add this scripting ability, you must first install the foo_spider_monkey_panel component. You can download it from the foobar2000 components page and install it by going to File > Preferences > Components > Install.
Locate your device names. In foobar2000, go to File > Playback > Device. Note the exact names of the output devices you want to switch between. They will be listed under the DS (DirectSound), WASAPI, or ASIO sub-menus, depending on your setup and installed components. For example: DS: Speakers (Realtek(R) Audio).
Scripting with the Spider Monkey Panel
Add a new Spider Monkey Panel: Right-click on your foobar2000 layout and select Add new panel > Spider Monkey Panel.
Open the panel configuration: Right-click the new panel and select Configure panel.
Insert the script: In the script editor window, enter the following code. This example shows a simple script to cycle between two devices, triggered by a click.

// Configuration
var devices = [
"Playback/Device/Realtek Digital Output (Realtek USB Audio) [exclusive]",
"Playback/Device/Headphones (USB-C to 3.5mm Headphone Jack Adapter) [exclusive]",
//"Playback/Device/Primary Sound Driver",
];
var currentDeviceIndex = 0;

// A label to display and click
function on_paint(gr) {
gr.FillSolidRect(0, 0, this.width, this.height, 0); // Black background
gr.DrawString(
"Switch Device",
gdi.Font("Arial", 14),
0xFFFFFFFF,
10,
10,
this.width,
this.height
);
}

// A function to handle the click and change the device
function on_mouse_lbtn_up(x, y) {
currentDeviceIndex = (currentDeviceIndex + 1) % devices.length;
var deviceName = devices[currentDeviceIndex];
fb.RunMainMenuCommand(deviceName);

// Optional: Show a popup to confirm the device change
//fb.ShowPopupMessage("Switched to: " + deviceName.split("/").pop(), "Device Switch");
}

Every song started with the volume dial on the speakers set to 0 or near 0. I've adjusted loudness for every song to where it felt good - this was different for every song and every trial. Volume settings were very close and I could not guess the source from fumbling with the volume dial.

Results:
Out of 30 trials, Mojo won on 18 trials. 10 trials were a draw. Apple won 2 trials.

• Chi-square statistic: 12.8
• p-value: 0.00166
• The p-value (0.00166) is much smaller than the typical significance level of 0.05.
• This means I reject the null hypothesis and conclude that there is a statistically significant difference in the preference between Mojo and Apple Dongle.

Here are the scores for 30 trials (flat = flat RCA switch position, this turned out to correspond to the Apple dongle; Sticking = the switch was sticking out, this turned out to be Mojo): https://i.postimg.cc/XY65mqLy/Mojo-experiment.jpg

Discussion:
It was hard to find differences on all tracks, there were no "night and day" tracks. Acoustical instruments, vocals and acoustic drums sound "more flowing" on Mojo. Electronic music sounded extremely similar, although maybe I was still getting accustomed to the test.

Shortcomings:
1).I always knew that each trial had 2 different sources, this made things easier - this is a biggie, and I still hope to convince my partner to donate 30 minutes of her life to redo this.
2).Although the volume dial settings were very similar for both sources (about the 1st mark on the drawn scale- depending on the song), and I did not look at this scale when setting volume, I may have developed a subconscious feel for which source corresponded to which volume setting. Asking my partner to go louder or quieter would also mitigate this. I don't want to rely on precise volume matching as changing volume for each song and genre is a natural way to listen to music and I tend to doubt the precision of those measurements.

Just wanted to applaud the effort you put into this test! Really nice work and well thought out!
The shortcomings you identified yourself could possibly influence the end result, but you seem to already recognize this.

That said, volume matching the DAC outputs would make the test more robust.
My preferred way of level-matching is to measure output level of both DACs with REW using a test tone (e.g. 1kHz sine) and then matching DAC output levels to <0.1dB precision. I'll usually measure the level with each DAC in the full signal path to account for potential differences when loaded, and use an ADC with a high input impedance (connected in parallel with the load).

Level-matching with REW makes it easy to also do basic sweep measurements, which is a nice way to detect potential differences in frequency response or THD (which can come in handy later when interpreting listening test results).

 

[ObjectiveSubjectivist]

So it was not level matched right?

 

[DanielT]

So it was not level matched right?

TS: The experimenter chooses if it is played from different sources (e.g., dongle then Mojo) or on the same source (eg., dongle twice). Volume is set to zero in the beginning of each song, the participant is free to adjust it.

But:

There could absolutely be systemic bias, because the dongle starts out at a much lower volume than the Mojo, meaning your volume tuning is influenced by the anchoring effect. As pointed out by other members, level matching is really, really important.

 

[ObjectiveSubjectivist]

TS: The experimenter chooses if it is played from different sources (e.g., dongle then Mojo) or on the same source (eg., dongle twice). Volume is set to zero in the beginning of each song, the participant is free to adjust it.

But:

Right, so Im not drunk