Suggestion: Audio interface shoot-out!

[ALMS]

Hi ASR,
I've just bought my self a new headphone amp (Aune x7s) and have been looking at which dac to get along with it. Now, since I'm making music myself and have a condenser microphone, the thought of using an audio interface as a dac sprung to mind. Now the issue is that the actual sound quality of the dacs in these aren't really being tested for, it's more of a question of features and adc quality. - Right, so I was thinking that it'd be really cool to get actual measurements of the sound quality of a couple of popular model-series. Examples of the models in question:
Series:
- Focusrite Scarlett - series
- Steinberg UR - series
- Audient iD - series
- Presonus AudioBox-series

Models without series:
- Native instruments komplete audio 6

Beyond testing each SKU against their respective competitor, I find that it begs the question as to whether or not the sound quality changes within each of the product series. Furthermore - which model is the absolute king in audio quality and which is the "sweet-spot" in price to performance.
I hope it's an idea that worthwhile for you to peruse and thank you for making such great and substantial content!

Best regards,
ALMS

 

[RayDunzl]

actual sound quality

How do you measure that?

 

[ALMS]

How do you measure that?

Ugh, the measured performance tests which amirm makes use of?

 

[Blumlein 88]

Welcome to ASR ALMS as this is your first post.

I don't know if Amir will make a project of this. And I believe so far Amir has only tested the DAC portions.

I could contribute testing of a Scarlett 18i20 as I have one. I have some other ADCs that surpass its measured quality so I could show what the 18i20 can do as a DAC.

BTW, of those you list, I would likely go with the Audient ID14. It would be nice to have good measures of them all in that we agree.

 

[amirm]

Welcome to the forum. I would love to test products in this category. Problem I have is that they are too expensive for me to buy. If members can loan them to me I would be happy to do the work.

 

[March Audio]

I can provide measurements of my MOTU 8A (already have elsewhere). Otherwise in terms of subjective testing that will always be a difficult issue. Only practical way I see is a multiple replay / re-record loops and see which remains closest to the original track.

It annoys me that there is little in the way of measurement data for Audio interfaces out there

 

[Blumlein 88]

I can provide measurements of my MOTU 8A (already have elsewhere). Otherwise in terms of subjective testing that will always be a difficult issue. Only practical way I see is a multiple replay / re-record loops and see which remains closest to the original track.

It annoys me that there is little in the way of measurement data for Audio interfaces out there

Gives me an idea for a fun thread or I think it might be. If we could pick one or a few widely available (or freely downloadable) tracks that were very well recorded, then loop them thru what interfaces our membership has. Make the files available so comparisons could be done. Say at least one and four loops. I would say more, but that can get onerous to do. I think many bits of kit might be subjectively indistinguishable even at 4 loops.

 

[bennetng]

Gives me an idea for a fun thread or I think it might be. If we could pick one or a few widely available (or freely downloadable) tracks that were very well recorded, then loop them thru what interfaces our membership has. Make the files available so comparisons could be done. Say at least one and four loops. I would say more, but that can get onerous to do. I think many bits of kit might be subjectively indistinguishable even at 4 loops.

Guess you may be interested in reading this thread:
https://www.gearslutz.com/board/gea...s/862592-lynx-hilo-vs-prism-orpheus-lyra.html

Also, about your proposal, a potential flaw in performing such tests are channel balance and level matching. For example a 0.1dB mismatch in L/R channel or I/O level will be accumulated in multiple loops and reveal the answer in an unfair way, it is not easy to adjust pots and faders on physical hardware with great precision, despite what the so-called "unity" gain is.

Very careful and precise adjustment can be performed in DAWs in float32/64 mode and requantize to 24 bit for repeated loopback tests.

Intersample peak is a potential problem as well, it can be solved by using some quieter reference tracks.

 

[amirm]

Also, about your proposal, a potential flaw in performing such tests are channel balance and level matching. For example a 0.1dB mismatch in L/R channel or I/O level will be accumulated in multiple loops and reveal the answer in an unfair way, it is not easy to adjust pots and faders on physical hardware with great precision, despite what the so-called "unity" gain is.

That's a good point. One way to get around it is to switch the channels on each run. Probably would be best to run the test in mono.

 

[Blumlein 88]

Guess you may be interested in reading this thread:
https://www.gearslutz.com/board/gea...s/862592-lynx-hilo-vs-prism-orpheus-lyra.html

Also, about your proposal, a potential flaw in performing such tests are channel balance and level matching. For example a 0.1dB mismatch in L/R channel or I/O level will be accumulated in multiple loops and reveal the answer in an unfair way, it is not easy to adjust pots and faders on physical hardware with great precision, despite what the so-called "unity" gain is.

Very careful and precise adjustment can be performed in DAWs in float32/64 mode and requantize to 24 bit for repeated loopback tests.

Intersample peak is a potential problem as well, it can be solved by using some quieter reference tracks.

I have done this before, had a thread about some 8th gen copies here at one time. I don't think the files are up for download anymore.

Did 8 runs. I attached a full scale tone at the front of each file to set levels. One of my devices at the time worked in discrete 1 db steps. So I set the first run carefully so it was less than 1 db below full scale. After a couple loops I would add 1 db gain. Didn't run into any intersample over problems. Though such a thing would be possible. When I was done I think the level difference was around 1/3 db. I don't recall any appreciable channel imbalance. I did touch up the levels to a precise match using software running in 32 bit float. Did each channel individually so it would have fixed any channel imbalance that existed.

I also did that once with a few different DACs. I forget, but that was probably only one loop. A couple of good DACs, a laptop on-board sound card, and a smartphone (and the original file). Let people choose their preferred DAC without knowing which was which. None were really obvious except the smartphone. John Swenson chose it because he preferred the sound. It did have a severe roll off starting near 14-15 khz and being 10 db down at 20 khz.

 

[ALMS]

Welcome

Welcome

Thank you guys!

I could contribute testing of a Scarlett 18i20 as I have one. I have some other ADCs that surpass its measured quality so I could show what the 18i20 can do as a DAC.

BTW, of those you list, I would likely go with the Audient ID14. It would be nice to have good measures of them all in that we agree.

Would be kinda cool, if it turned into a community project like that! The Audient's definetly have my fancy, though I'm not sure how to pick between the iD4 and iD14. Both are supposed to have great audio quality, though spec-wise the dacs in each have pros and cons relative to eachother. Not sure what i amounts to in the end.

Problem I have is that they are too expensive for me to buy. If members can loan them to me I would be happy to do the work.

Ah, I see. Sadly I'm based in the EU, so shipping would be a hindrance for me to do that. A solution could be to make a patreon or some other means of crowdfunding, when projects like this one comes up. Then you'd even be able to sell off the gear afterwards, if need be.

Gives me an idea for a fun thread or I think it might be. If we could pick one or a few widely available (or freely downloadable) tracks that were very well recorded, then loop them thru what interfaces our membership has. Make the files available so comparisons could be done. Say at least one and four loops. I would say more, but that can get onerous to do. I think many bits of kit might be subjectively indistinguishable even at 4 loops.

That's a great idea! It would require some smart heads to formulate an exact methodology, to make the results consistent though (as people already have picked up on ). For my sake though, I'm just interested in the quality of the dac portion of the devices and how they perform in relation to dedicated dacs!

 

[Purité Audio]

I have all the Audient pres here they are surplus to my requirements, send me a Pm if you might be interested in the id22 at a considerable discount.
Keith

 

[Blumlein 88]

@ALMS

https://www.audiosciencereview.com/...srite-18i20-dac-measurements.2128/#post-57514

In case you didn't see it, I have posted DAC measurements of the Focusrite 18i20 interface.