Audio stutters with USB Dacs on Macbook M1 Pro

[khensu]

Considering it only happens under heavy memory usage, I’d say it’s very likely the problem is software related, or at the very least, could be solved in software.

 

[Zensō]

The Qudelix 5K plays music fine, as do my old Harman Kardon Soundsticks. And Elements/SoundSource seems to help by adjusting the buffer. Maybe a software solve is be possible?


After reading about the HDMI option, I had a quick google and searched the forum, but I'm guessing there's no easy way to extract audio from HDMI to a USB DAC? Cos all the HDMI splitters I see don't have USB out.

The Quedelix is one of the few DACs that is UAC1, which makes me wonder if the JDS Labs Element III might work because it’s switchable between UAC1 and UAC2. I’m tempted to buy one just to try it out.

Some USB DACs also have an optical input.

 

[Zensō]

Considering it only happens under heavy memory usage, I’d say it’s very likely the problem is software related, or at the very least, could be solved in software.

That makes sense.

 

[Axo1989]

I've been communicating with Apple about this issue for more than a year by now. Realistically, you should ask yourself: Is there any reason to believe that the stuttering will be fixed in the future? After all, it was not fixed within one year. That‘s a long time compared to the paxe at which Apple releases new futures or fixes other bugs. Therefore, I don't think it is ever gonna happen.

I suspect that one of these two things is going on:

(1) It's not fixed because they cannot fix it. It's a hardware problem and Apple tries to play it low-key with plausible deniability so that the audio stuttering does not end in a class action lawsuit. They might be able to reduce the stuttering, but it will never be fixed.

(2) It's a software problem, but really low on their priority list. Realistically, this also means that it won't be fixed. Apple has much more glaring issue to fix. For instance, Bluetooth is an absolute mess even when only using their own products such as AirPods, Magic Keyboard and Trackpad. Facetime, hand over, and screen share have been around for quite some time, but they are also a complete mess. Instead of fixing things, Apple is rolling out new features and macOS stays in a constant beta status.

It is extremely disappointing that we cannot play audio reliably with machines that have cost some of us upward of of USD 5k.

I don't think anything is gonna happen if we are not very vocal.

You must have different meanings for "absolute" and "complete" in conjunction with "mess" in that fourth paragraph. But I agree with the final sentences.

 

[Zensō]

I don't think anything is gonna happen if we are not very vocal.

I wonder what percentage of people on M1 Macs are using external USB DACs? Seeing how most of the discussion on this issue is on audio forums, but very little can be found on MacRumors or the Apple Support forum, I’m guessing a fairly small number. Most are probably listening on Airpods, HomePods, or some sort of Bluetooth speaker, none of which are exhibiting this problem. That might explain the lack of a solution after a year or more.

 

[theREALdotnet]

I wonder what percentage of people on M1 Macs are using external USB DACs? Seeing how most of the discussion on this issue is on audio forums, but very little can be found on MacRumors or the Apple Support forum, I’m guessing a fairly small number. Most are probably listening on Airpods, HomePods, or some sort of Bluetooth speaker, none of which are exhibiting this problem. That might explain the lack of a solution after a year or more.

Good point. I suppose in order to prevent any support cases with Apple from getting put on the “third party device issue” track, it would be useful to include Apple’s own USB DAC (the USB to 3.5mm adapter) in the test setup.

Has anyone observed stuttering with that device?

 

[Zensō]

Good point. I suppose in order to prevent any support cases with Apple from getting put on the “third party device issue” track, it would be useful to include Apple’s own USB DAC (the USB to 3.5mm adapter) in the test setup.

Has anyone observed stuttering with that device?

I’ll test this tomorrow.

 

[gochristoph]

I think everybody can reproduce the issue even with the internal loudspeakers. You have to go extreme with memory pressure and go for the highest sampling rate of 96 kHz in "Audio Midi Setup". This way I can reproduce the issue without any reliance on external/third-party equipment.

It's far less prominent than via all USB DACs and Bluetooth devices that I tried and audible more as a crackle/pop (a very short stutter).

Did anybody with a M1 Max or M2 Max experience the stuttering? Memory bandwidth is higher there (400GB/s vs 200GB/s). Perhaps that mitigates the issue.

 

[pengin15]

Considering it only happens under heavy memory usage, I’d say it’s very likely the problem is software related, or at the very least, could be solved in software.

This is what I'm thinking too - more about how Apple implements its audio driver than an actual hardware defect (Christ I really hope this isn't the case, that would be the most annoying thing) since console seems to indicate that coreaudio just...dies whenever memory load gets too high. I'd imagine something can be done since Windows's audio driver doesn't really panic with RAM usage (moreso with CPU, which is understandable) and Intel Mac doesn't have this issue at all.

It bears remembering discussions in the previous few pages that professional audio software doesn't seem to encounter this issue at all which, if that is really the case, implies that there's something different with how coreaudio works when those professional audio software are running which theoretically should be implementable for us on the consumer side.

 

[vurt]

Did anybody with a M1 Max or M2 Max experience the stuttering? Memory bandwidth is higher there (400GB/s vs 200GB/s). Perhaps that mitigates the issue.

I'm in touch with a Questyle rep who told me he's not experiencing it on an M1 Pro Max.

 

[antcollinet]

I'm in touch with a Questyle rep who told me he's not experiencing it on an M1 Pro Max.

View attachment 292432

But I am also not experiencing it on an M2pro.

Though my DAC is old - is it also related to the version of USB protocol?

 

[Zensō]

But I am also not experiencing it on an M2pro.

Though my DAC is old - is it also related to the version of USB protocol?

It appears older UAC1 DACs are not affected.

 

[vurt]

Early on in the thread there was some attempt to test the UAC difference. I don't think there was anything conclusive.

 

[vurt]

But I am also not experiencing it on an M2pro.

Though my DAC is old - is it also related to the version of USB protocol?

Wonder what's changed between M1 Pro and M2 Pro. Is it memory bandwidth or something?

Urgh what other alternatives are there? Listen to music off the phone or a music player?

 

[Zensō]

Early on in the thread there was some attempt to test the UAC difference. I don't think there was anything conclusive.

Thanks. There have been so many false starts and fixes that later failed I’m starting to lose track.

 

[Old Listener]

The Quedelix is one of the few DACs that is UAC1, which makes me wonder if the JDS Labs Element III might work because it’s switchable between UAC1 and UAC2. I’m tempted to buy one just to try it out.

I use a UAC1 DAC in my desktop system with a 16GB M1 Mac mini. I've never heard an audio stutter while using that system.

In my main system (16GB M1 MacBook Pro), I use a DAC that is switchable between UAC1 and UAC2. After thinking that I had heard a very brief stutter twice, I switched the DAC to UAC1 and haven't heard a stutter since.

I don't use any heavyweight apps but memory use is often above 13GB. I have lots of Safari tabs open displaying text pages with tons of ads including one or more video ads in each tab.

 

[Trell]

I have lots of Safari tabs open displaying text pages with tons of ads including one or more video ads in each tab.

Oh my! Perhaps add an add blocker and only allow those sites you want to support?

 

[Zensō]

Urgh what other alternatives are there? Listen to music off the phone or a music player?

The HDMI extractor fix seems good, if not a bit of a kludge.

The software solutions to increase the audio buffer (Soundsource, Element, Black Hole) improve the situation but increase latency and are not 100% effective.

The M2 seems immune, which is why I’m still wondering if there’s an issue with the M1 chip.

Other than that, anything that eliminates a USB 2 or newer DAC from the chain seems to pretty much solve it. Personally, I’ve gone to using my studio monitors only for production and critical listening when I’m not multitasking, and a pair of HomePods for casual listening when I am multitasking and don’t need the highest fidelity to enjoy the music.

I’m not a conspiracy theorist by any measure, but if one was so inclined they could easily imagine Apple doing this intentionally to sell more HomePods, AirPods, etc.

 

[pengin15]

Do also follow this discussion thread at Apple's user forum:

Macbook pro 16" M1 pro 2021 popping sound… - Apple Community

discussions.apple.com

This might be a manifestation of the same issue these folks have been facing, and it looks as if they were caused by the same coreaudio issue.

Here's a reddit post identifying their issue and identifying the suggestion to be - you guessed it - just don't tax the RAM too much:

Now for the bad news - Apple users have been identifying this issue for years now since the M1 and there doesn't seem to be any solution in sight since it's a low-level bug that is not easy to fix, and they're a far bigger crowd than us...

EDIT:
Given the above, it's also possible that this bug might have been present since the Intel days, but was not an issue previously since the Intel machines does not use the current unified memory system in which the same memory bandwidth now have to be shared between RAM and VRAM, meaning that 16GB on Apple Silicon gives you significantly less headroom than 16GB of RAM in the Intel machines would.

If M2 really did fix it through hardware (and given the relative lack of anecdotal evidence so far, I doubt it's really fixed by M2), that means the only solution is to pony up $4000 for a machine that runs less efficient and hotter than our current M1 devices. This is quite annoying.

I do wonder if this could have been fixed by Apple being more aggressive with using swap and if maybe, just maybe, macOS Sonoma is better with this issue so far since Apple is just doing that.

 

[antcollinet]

Thanks. There have been so many false starts and fixes that later failed I’m starting to lose track.

So I think the device I am currently using (soundblaster omni 5.1 - actually being used as USB to Optical) is UAC1.

When I get around to it, I'll go directly USB into my MiniDSP flex (which I'm assuming is UAC2) and see what happens.