Low latency audio interface for playing vsti's live

[sleepwalk]

Hi folks, I'm back at the keyboard corner, because this is the best place there is to ask the question I have. I just got a new laptop! It's a 17 inch Lenovo with the I7-9750H processor, 16 GB ram, and a 1TB SSD drive. It has USB and USB3, but it does not have Thunderbolt. I am excited to use it to play some of my favorite vst instruments, such as, Piano in Blue, VB3, Scarbee, various analog synth emulations, various Kontakt libraries, etc..

 

I have been trying it with ASIO4ALL but there is some latency. In Ableton Live it says 6.65 seconds round trip, something like that, but I don't think I'm really getting it that low...I can feel the latency, no question...

 

So yes, I need an audio interface! I am trying to keep it under $300. If it only has 1 or 2 inputs that is okay. I'm not really going to be recording with it, I have other gear for that. So far I am thinking of an RME Babyface (used), Zoom UAC-2, or maybe one of the Scarlett interfaces? I am trying to get LOWWWW latency -- I don't want to feel any lagging between when I hit the key and the sound comes out, at all....is that possible? Is it possible for it to feel as responsive as a hardware keyboard? Because that is what I want.

 

One other question: I have a midi to usb cable that I connect directly to my laptop. Does it make an difference using an interface with a MIDI in? Would that help in regard to latency? Or is all the latency we hear a result of audio processing?

 

Any suggestions are appreciated!

 

Thanks -Jon

[SteveUK]

Latency is largely due to audio processing.

 

RME is considered one of the best for low latency drivers, and for continued support for their products.

 

I use a Radial Key Largo which I find gives acceptable latency and serves as a sub mixer. May be overkill for what you need though

[Al Coda]

Hi folks, I'm back at the keyboard corner, because this is the best place there is to ask the question I have. I just got a new laptop! It's a 17 inch Lenovo with the I7-9750H processor, 16 GB ram, and a 1TB SSD drive. It has USB and USB3, but it does not have Thunderbolt. I am excited to use it to play some of my favorite vst instruments, such as, Piano in Blue, VB3, Scarbee, various analog synth emulations, various Kontakt libraries, etc..

 

I have been trying it with ASIO4ALL but there is some latency. In Ableton Live it says 6.65 seconds round trip, something like that, but I don't think I'm really getting it that low...I can feel the latency, no question...

 

You mean 6.65ms for sure !

 

I see 5.80ms, 44.100 KHz, 256 buffer size w/ ASIO4ALL and (Topten Software) Cantabile Lite 3.0 build 3627 x64, i7 4790k, AsRock Z97 Extreme6.

I see the same w/ Reaper 6 and Reason 11.1 ...

 

In ASIO4ALL Control Panel,- you have to set the buffersize lower than it´s default setting of 512samples AND set "buffer offset" to zero (0) instead leaving the default of 4ms !

 

I´m pretty sure I can go lower than 256 samples buffersize,- but that´s actually my test setup w/ the Realtek HD Audio codec.

 

I also have RME PCI and S|C XITE-1 (DSP farm) PCIe based cards and can go even lower.

 

So yes, I need an audio interface! I am trying to keep it under $300.

 

I really doubt it will be better that 6.65ms roundtip w/ cheapo USB interfaces.

To come down w/ latency you need PCIe based interfaces,- or for laptop,- Thunderbolt,- because it IS PCIe !

Most USB3 compatible interfaces are still USB2 in realworld.

There might be exceptions, but I never tried.

With USB, there´s always a controller in between (in the way) while PCIe is connected directly to the CPU, especially on some motherbords when the card is inserted in one of the x16 graphics card slots which offer highest bandwidth always.

There´s no prob inserting a PCIe x1 into a PCIe x16,- but you cannot do that w/ a laptop.

 

But when you don´t have Thunderbolt available on your laptop and are forced to use USB,- a RME Babyface or any other RME supporting USB is possibly the best solution.

 

 

One other question: I have a midi to usb cable that I connect directly to my laptop. Does it make an difference using an interface with a MIDI in? Would that help in regard to latency? Or is all the latency we hear a result of audio processing?

 

USB for ingoing MIDI is o.k., for outgoing MIDI to external hardware modules, not so much.

 

AFAIK, the only tight one for outgoing MIDI is a single DinMIDI port Expert Sleepers unit.

 

A.C.

[BluMunk]

Have you considered siting closer to your speakers?

[bfields]

I have been trying it with ASIO4ALL but there is some latency. In Ableton Live it says 6.65 seconds round trip, something like that, but I don't think I'm really getting it that low...I can feel the latency, no question...

 

So yes, I need an audio interface! I am trying to keep it under $300. If it only has 1 or 2 inputs that is okay. I'm not really going to be recording with it, I have other gear for that. So far I am thinking of an RME Babyface (used), Zoom UAC-2, or maybe one of the Scarlett interfaces? I am trying to get LOWWWW latency -- I don't want to feel any lagging between when I hit the key and the sound comes out, at all....is that possible? Is it possible for it to feel as responsive as a hardware keyboard? Because that is what I want.

 

One other question: I have a midi to usb cable that I connect directly to my laptop. Does it make an difference using an interface with a MIDI in? Would that help in regard to latency? Or is all the latency we hear a result of audio processing?

 

You can measure total roundtrip audio latency using a procedure like they describe in the audacity manual: https://manual.audacityteam.org/man/latency_test.html

 

The only time I've measured it carefully, I found that the built-in audio on my laptop had better latency than anything else I tried. Which makes sense to me--the typical built-in audio is probably right on the PCI bus, isn't it?

 

You could do similar measurements to compare hardware vs software latency: if you have a hardware synth that you can set to both play internal sounds and transmit MIDI, do that, set both to sounds with a fast attack, record the result, and measure the distance between the two attacks in an audio editor.

 

Have you considered sitting closer to your speakers?

 

Yep, note sound travels at about a foot a millisecond, so a few feet normally won't matter, more may. But, yes, if you're trying to get precise latency measurements or something then you need to take it into account.

[sleepwalk]

You mean 6.65ms for sure !

 

I see 5.80ms, 44.100 KHz, 256 buffer size w/ ASIO4ALL and (Topten Software) Cantabile Lite 3.0 build 3627 x64, i7 4790k, AsRock Z97 Extreme6.

I see the same w/ Reaper 6 and Reason 11.1 ...

 

In ASIO4ALL Control Panel,- you have to set the buffersize lower than it´s default setting of 512samples AND set "buffer offset" to zero (0) instead leaving the default of 4ms !

 

I´m pretty sure I can go lower than 256 samples buffersize,- but that´s actually my test setup w/ the Realtek HD Audio codec.

 

I also have RME PCI and S|C XITE-1 (DSP farm) PCIe based cards and can go even lower.

 

I followed these instructions, and the best I can do in Ableton Live testing with VB3 (with buffer offset at 0, and ASIO Buffer Size at 104 samples) is 6.71 ms overall latency. This is with ASIO4ALL at sample rate 44100.

 

Is 6.71 considered good? Because for me, this is not really acceptable for any real live playing. I have the midi keyboard running a second midi out into a Roland XV-2020 sound module at the same time (playing an organ sound), and I have a set of headphones on each, the Roland, and the laptop. I go back and forth with the headphones of each, and the feel is noticeably more precise on the Roland. For any kind of fast run on the keyboard with the organ sound, I can basically hear the VB3 sound playing after I hit the key. At least it feels that way in comparison to the Roland.

 

Would a good USB interface improve on this? I was under the impression that the technology was at a point where it had reached a standard where a usb laptop is playable in a live setting. I can say that it would be okay for synth pads or something like that, but that would be about all.

 

I have a USB-C input on my laptop. Is there an interface that can utilize the greater speed of USB-C to lower the latency even more?

 

What about an external SSD? Would that help? Right now I am running everything off the single SSD.

 

As I said I am monitoring through headphones so my distance from speakers is not a factor.

 

I bought the laptop from Best Buy on a Black Friday deal. Do I need to return it and get a thunderbolt laptop instead, or will I be able to work with this one? Everything I had read lead me to believe that the USB technology had improved so much that latency was no longer an issue.

 

Thanks to all those replying, any advice is appreciated!

[bfields]

Is 6.71 considered good? Because for me, this is not really acceptable for any real live playing.

 

Yes. If the total delay is really that noticeable, then it's a lot more than 7ms. So the problem is somewhere else, I don't know where.

 

I have the midi keyboard running a second midi out into a Roland XV-2020 sound module at the same time (playing an organ sound), and I have a set of headphones on each, the Roland, and the laptop. I go back and forth with the headphones of each, and the feel is noticeably more precise on the Roland. For any kind of fast run on the keyboard with the organ sound, I can basically hear the VB3 sound playing after I hit the key. At least it feels that way in comparison to the Roland.

 

Can you mix the two or otherwise arrange to be able to hear both at once? It might be interesting to know whether you can actually hear a delay between the two triggering.

 

Would a good USB interface improve on this? I was under the impression that the technology was at a point where it had reached a standard where a usb laptop is playable in a live setting.

 

Yes, sounds like there's something wrong with your setup. I doubt a USB audio interface is what you need, but maybe somebody else has troubleshooting ideas.

[sleepwalk]

 

Can you mix the two or otherwise arrange to be able to hear both at once? It might be interesting to know whether you can actually hear a delay between the two triggering.

 

 

So while you were replying, I was running this very test. I ran the audio out from the Roland and the audio out from the laptop both to my Zoom R16 (digital multitrack recorder). Then I loaded the files into Ableton and looked at them. The difference averages about 6 ms (laptop lagging behind). I just did organ single note hits. The difference was sometimes 5 ms, sometimes 7 ms. The average was around 6 ms.

 

There is also MIDI latency to consider, but I am not going to bother with that. If the laptop can respond with the same feel as a sound module I would be pleased.

 

Maybe I am just one of those people that notices the latency more than others? Or am I overthinking it? If I didn't know that there was latency, would I still feel it? I mean, all my little test does is confirm what the ASIO4ALL was telling me. People say that 6-7 ms is not noticeable.....but some people swear that they notice it. So how low can we get it? Zoom says says theirs can do 2.2 seconds. Is that legit? Maybe I would not notice it then.

[bfields]

So while you were replying, I was running this very test. I ran the audio out from the Roland and the audio out from the laptop both to my Zoom R16 (digital multitrack recorder). Then I loaded the files into Ableton and looked at them. The difference averages about 6 ms (laptop lagging behind). I just did organ single note hits. The difference was sometimes 5 ms, sometimes 7 ms. The average was around 6 ms.

 

OK, I stand corrected, then! And googling around.... I guess you're not alone.

 

But with the speed of sound about 1ft/ms, that's the equivalent of monitoring with a speaker six feet away, which feels like something a person should be able to deal with, as long as there's not a lot of variation in the latency.

[Markay]

Easy way to test whether you are as sensistive to latency as you suggest is to play a hardware board, then move the speaker 6' further away. If you can notice the difference then maybe you are hyper sensitive and can feel latency of around 6ms. From Billy Joel down there are many artists successfully playing VST's live where midi travels along long snakes to off stage computers and then audio back to their onstage monitors. There is way more than 6ms latency in those rigs.

 

From the feel you describe there has to be more than the reported 6 ms of latency. One way you can try and measure it is to set up an audio recorder, like a Zoom, put some thimbles on your finger and hit a key hard enough that the thimble makes an audible noise. Then open the recorded wav file in Audacity and see how many ms elapse between the the sound of the thimble and rise from the leading edge of the VST sound. Deduct 1ms for each foot that the speaker is from the recorder.

[sleepwalk]

I did a couple more tests:

 

Recorded into Zoom multitrack recorder:

 

Direct audio from Roland sound module vs. SM58 on keystrike audio = Roland audio average 16 ms behind

Direct audio from laptop vs SM58 on keystrike audio = Laptop audio average 30 ms behind

 

This means on the Roland it's about 16 ms latency which is MIDI latency plus whatever additional processing time. ?

Not sure why the difference between the laptop and the Roland was 14 ms in this test, when it was 6-7 ms in the other test. Different test method with different factors I guess?

 

Either way, would these numbers be considered normal?

I just have a generic MIDI to USB cable going into the laptop. Is there a better way? Is there MIDI to USB-C? What interface can I use to improve this latency?

[TomKittel]

 

Any suggestions are appreciated!

 

Thanks -Jon

 

I am using an Audient ID14:

 

https://www.thomann.de/gb/audient_id14.htm?ref=intl&shp=eyJjb3VudHJ5IjoiZ2IiLCJjdXJyZW5jeSI6IjIiLCJsYW5ndWFnZSI6ImVuIn0%3D

 

Very low latency, excellent sound quality. The BurrBrown converters are significantly better compared to it's competition in it's price range. Much cheaper than the RME Babyface which I also owned. There is no perceptible difference in latency or sound to the RME.

 

JMTC

 

[Markyboard]

 

Either way, would these numbers be considered normal?

I just have a generic MIDI to USB cable going into the laptop. Is there a better way? Is there MIDI to USB-C? What interface can I use to improve this latency?

 

This may give you some insight into the complexity of measuring and comparing latency numbers:

 

DAW Bench

[Markay]

I did a couple more tests:

 

Recorded into Zoom multitrack recorder:

 

Direct audio from Roland sound module vs. SM58 on keystrike audio = Roland audio average 16 ms behind

Direct audio from laptop vs SM58 on keystrike audio = Laptop audio average 30 ms behind

 

This means on the Roland it's about 16 ms latency which is MIDI latency plus whatever additional processing time. ?

Not sure why the difference between the laptop and the Roland was 14 ms in this test, when it was 6-7 ms in the other test. Different test method with different factors I guess?

 

Either way, would these numbers be considered normal?

I just have a generic MIDI to USB cable going into the laptop. Is there a better way? Is there MIDI to USB-C? What interface can I use to improve this latency?

 

I would not consider that normal. 30 ms is high and definitely in the range that I think most would notice. The cables are not the issue, midi involves the transfer of a small number of bits which USB can handle just fine. Similarly virtually all audio interfaces are USB 2 and perform well with low latency. The issue lies with drivers, hardware, or settings on your machine. There are a number of good guides on how to configure a PC for best audio performance. . For example are you running max power mode?

 

There is a good guide available on the Cantabile site see this link Glitch Free

[Markay]

 

Either way, would these numbers be considered normal?

I just have a generic MIDI to USB cable going into the laptop. Is there a better way? Is there MIDI to USB-C? What interface can I use to improve this latency?

 

This may give you some insight into the complexity of measuring and comparing latency numbers:

 

DAW Bench

 

Agreed but the finger to ear test is what matters. If we can't feel any latency it doesn't matter what it is. As I understand it 10ms is about the threshold for most people, higher for some. Vin's work is great great for ranking drivers and interfaces. Also what matters playing VI's is midi in > VI > DAC audio out latency, not audio in ADC > VI > DAC audio out which will always have higher latency.

 

[Markyboard]

Yup- we've discussed this at length before. I think somewhere in the text portion it addresses this. Independent of application it's still of some use for comparative purposes.

[Theo Verelst]

This sounds like science. There are a few things to consider when talking about practical latency problems, first of all, a main problem is the CPU in the computer itself, taking part in operating system related activities is made to do things unlike it used to be in the microprocessor time, namely a lot of reasonably course grain pipe-lining makes it as fast as it can be. So fast real time responses aren't the main dish for the modern Intel compatibles, the heat is more in supercomputing based computing with long crunch time. The main factors that decide on latency outside the actual few computations required for instance for a sample playback program are contention with other programs and OS tasks which keep processors/threads occupied, and the time it takes for a thread processor to do your usb midi or sample reading time, the contention and strategy execution connected with resource management like drives, Usb hubs, CPU communication ports. The there's the memory and caching structure which has to deal with competing users like the per thread pre-fetchers bandwidth hunger, the memory controller access contention from all processors, the access time for random memory locations being way less favourable then streaming contiguous locations, and very important: the virtual memory management and it's segment/page bookkeeping, both in the OS and the virtual to physical translation look aside buffer from the CPU.

 

In terms of how long latency is acceptable there a lot of talk possible connected with the following subjects: actual accuracy versus average latency, the type of sound you're playing, the type of DAC management that takes place. It can be argued that a good musician playing through an analogue delay line in a Hall can probably get used to pretty long delay lines, like an example can be a pipe organ player. If the instrument responds predictably and nice, even a tenth of a second can be ok to play songs without utter discomfort. If however you have to play a funky lick on a machine that tries to fill your sample playback requests by feeding the output sound buffer as quick as possible, and without any natural sounding filtering components, even 5 mili seconds of delay can be quite annoying in an accurate setting, and while hearing other instruments, regardless of a bit of distance between the monitors and the musicians(s). It's the "hole" in the response that works with the always-the-same reconstruction errors of most (pretty much all) digital to analog converters which makes it very hard to "sync" your music playing with the always important natural acoustics. Which brings us to the final important idea: it's not just latency that's important, but the *constancy* of the latency is possibly more important. If you now for sure your organ pipes are going to sound exactly as you're used to after 30 meters of sound travel and even longer of waves building up in a church, that's a lot better than putting down your clavinet tone and having to gamble which phase the low frequency note are going to have when compared with the bass drum for instance. 10 milliseconds of uncertainty can completely kill the solo or bass line, at least for serious musician.

 

T

[PianoMan51]

Theo,

 

We"ve exchanged a few barbs over the years. But, seriously, whatever meds you"re on, keep it up. This is a 100% useful (IMHO) and important post.

 

Bravo!

 

Jack

[Theo Verelst]

I don't think it's ok to accuse people of using meds, not cool.

[PianoMan51]

Sorry. Let me restate:

 

'Whatever you"re doing in your life, keep it up!'

 

The original was meant as a jest, but I missed my mark. Please take the intention as as positive.

[Jazz+]

This sounds like science. There are a few things to consider when talking about practical latency problems, first of all, a main problem is the CPU in the computer itself, taking part in operating system related activities is made to do things unlike it used to be in the microprocessor time, namely a lot of reasonably course grain pipe-lining makes it as fast as it can be. So fast real time responses aren't the main dish for the modern Intel compatibles, the heat is more in supercomputing based computing with long crunch time. The main factors that decide on latency outside the actual few computations required for instance for a sample playback program are contention with other programs and OS tasks which keep processors/threads occupied, and the time it takes for a thread processor to do your usb midi or sample reading time, the contention and strategy execution connected with resource management like drives, Usb hubs, CPU communication ports. The there's the memory and caching structure which has to deal with competing users like the per thread pre-fetchers bandwidth hunger, the memory controller access contention from all processors, the access time for random memory locations being way less favourable then streaming contiguous locations, and very important: the virtual memory management and it's segment/page bookkeeping, both in the OS and the virtual to physical translation look aside buffer from the CPU.

 

In terms of how long latency is acceptable there a lot of talk possible connected with the following subjects: actual accuracy versus average latency, the type of sound you're playing, the type of DAC management that takes place. It can be argued that a good musician playing through an analogue delay line in a Hall can probably get used to pretty long delay lines, like an example can be a pipe organ player. If the instrument responds predictably and nice, even a tenth of a second can be ok to play songs without utter discomfort. If however you have to play a funky lick on a machine that tries to fill your sample playback requests by feeding the output sound buffer as quick as possible, and without any natural sounding filtering components, even 5 mili seconds of delay can be quite annoying in an accurate setting, and while hearing other instruments, regardless of a bit of distance between the monitors and the musicians(s). It's the "hole" in the response that works with the always-the-same reconstruction errors of most (pretty much all) digital to analog converters which makes it very hard to "sync" your music playing with the always important natural acoustics. Which brings us to the final important idea: it's not just latency that's important, but the *constancy* of the latency is possibly more important. If you now for sure your organ pipes are going to sound exactly as you're used to after 30 meters of sound travel and even longer of waves building up in a church, that's a lot better than putting down your clavinet tone and having to gamble which phase the low frequency note are going to have when compared with the bass drum for instance. 10 milliseconds of uncertainty can completely kill the solo or bass line, at least for serious musician.

 

T

I agree.

 

I cannot play piano VSTIs for latency reasons.

[Kurt W]

From Billy Joel down there are many artists successfully playing VST's live where midi travels along long snakes to off stage computers and then audio back to their onstage monitors. There is way more than 6ms latency in those rigs.

 

Well those cables shoulde be about 20 kilometers long to add about 1 millisecond of delay. Your ears must be exceptional :-)

Speed of light is 300 000 km per second, speed of a signal in a cable is about 2/3 of that. Sorry that should be more like 200 km.

[Theo Verelst]

Sorry. Let me restate:

 

'Whatever you"re doing in your life, keep it up!'

 

....

 

Thanks for that. I had one coffee today, no beer day today, no wine, not even fancy organic (actual) vitamins, so maybe some of the excellent coffee could be blamed..

 

T

[PianoMan51]

You are warmly welcome Theo. Here"s to happy holidays for you and yours.

[Throbert]

yep, Intel was never made for real time, but other than the CPU, the biggest determination of latency

is the quality of the driver. That's why RME reigns king their engineers not only write the drivers but

they also design the interfaces and firmwares. Here's more info from Vins radio show.

Hey All,

 

I popped this up on the main DAWbench thread, but I thought it would be worthwhile posting up the notice here as well.

 

With RME being a focus over the years for driver performance, I think many reading in will be interested in listening in to the chat with Matthias Carstens , who is the co-founder and lead developer at RME.

 

DAWbench Radio Show :

 

Episode 03: Audio Interface Low Latency Performance : RME â Setting the Bar !

 

Special Guest : Matthias Carstens

 

Run Time : 88 Minutes

 

DAWbench Radio Show : Here

 

Spotify : Here

Enjoy

 

peachh

 

 

 

[sleepwalk]

 

Any suggestions are appreciated!

 

Thanks -Jon

 

I am using an Audient ID14:

 

https://www.thomann.de/gb/audient_id14.htm?ref=intl&shp=eyJjb3VudHJ5IjoiZ2IiLCJjdXJyZW5jeSI6IjIiLCJsYW5ndWFnZSI6ImVuIn0%3D

 

Very low latency, excellent sound quality. The BurrBrown converters are significantly better compared to it's competition in it's price range. Much cheaper than the RME Babyface which I also owned. There is no perceptible difference in latency or sound to the RME.

 

JMTC

 

Thank you for the recommendation. I'll take a look at that one.

[sleepwalk]

Thanks for all the well-thought-out replies in this thread, a lot of good info here.

 

I downloaded Cantabile Lite to see if it would run my vstis any smoother than Ableton. It seemed a little better, a little less latency. But then I noticed, when I play a piano sound, with the pedal down, after a while, I'm getting a lot of pops and crackles. I thought with my new super-computer this kind of thing would be a non-issue. I will still try to find a good audio interface to see if this can be improved.

 

Maybe the laptop is better as a tool for recording, mixing and editing, but not so much for a live keyboard player. Maybe I should just get a sweet Nord for playing killer keys sounds.

 

I thought a lot of people on here gigged with laptops, and had good experiences. Am I wrong about that? I thought in 2019 this would be easy for a laptop. It just seems weird, they make these awesome sounds, but then I can't just play them, and go off live with them. Didn't people do that with Receptor? I think my laptop has more power than a Receptor, doesn't it?

[Markay]

Lots of folk here have successfully used laptops live for many years with no noticeable latency. I started out on Windows and switched to OSX because Apple build and sell their own DAW - Logic, VI's called AU's in the Apple World, and a VST Host, MainStage. To make it all work Apple also provide their CoreMidi and CoreAudio drivers that are tested and work. An OS that is built and tested to work for music. Music is not a priority for MS.

 

On Windows you to are using on OS on a wide range of hardware that can be bundled together in an infinite number of combinations. Some of these hardware combinations work brilliantly for music others don't - for no apparent reason.

 

I ended up paying the Apple tax because I want a system that just works out of the box.

 

My day job for the psst 30 years has existed because we manipulate MS OS's to do what is required for us to deliver our services. But once the sun goes down I prefer to play, not spend time debugging tech.

 

I am sure others will chime in with the brand and specs of a laptop running Windows that works for them.

 

But let me throw this out as a good starting point. Svengle "best Hackintosh laptops". What comes up will be those that have hardware nearly identical to that which Apple uses. Pick one of these, if it will run Mac OS with minimal changes you can be almost guaranteed it will also be good for music with Win10.

[Tusker]

Maybe the laptop is better as a tool for recording, mixing and editing, but not so much for a live keyboard player. Maybe I should just get a sweet Nord for playing killer keys sounds.

 

I thought a lot of people on here gigged with laptops, and had good experiences. Am I wrong about that?

 

Nords are sweet for sure, but today's laptops are eminently giggable. I've done countless gigs over six years though a complete newbie compared to Markay and some of the others here. I've used (various) Macs with no problems. Macs provide less hardware choices, so historically it's been quicker to bullet-proof a Mac rig, but the current range of PC alternatives are powerhouses. Generally, Mac laptops have had consistent low latency with the onboard audio while latency has been more variable when using on-board audio on the various PC laptops. The addition of a professional audio interface largely levels the playing field or tilts toward PC's on a price-performance basis. (imo) Latency should not be a problem. It will depend on how the elements are interacting (audio interface and drivers) in your rig ... and not on whether the laptop is powerful enough.

 

Based on what you have written, I am inferring that you are using the laptop's audio-out (and not a professional audio interface) for your current measurements. Is this correct? Get a good audio interface! Join the fun!

[Markay]

Tusker makes a good point about getting an audio interface. The benefits are it comes with an optimised audio driver, you shift audio processing off the laptop and most have master knob to alter volume on the fly. I use a 2nd Gen Focusrite. Around $120 when I got mine. Sure RME' are the Rolls Royce, but 5 or more times the expense for one way improvement of 2ms or so. You are searching for a 15 to 20ms improvement which, if audio is the issue here, any interface will fix.