Jim Roseberry

While it's on my mind...

Another example how monitoring thru software can open up new possibilities:

I've got one of the new HeadRush Gigboard guitar processors.

I've programmed a nice Marshall JCM-800 patch (with various optional boosts/EFX)... and I'd like to play/hear that in realtime with stereo Cab IRs (not just a single mono Cab).  HeadRush can run a pair of Cab IRs... but that (along with a single Amp) nearly maxes out its DSP.

With the ability to monitor thru software with 1ms total round-trip latency, I can set-up a stereo pair of Cab IRs in my DAW software.

Also, I can set-up pristine Reverb and Delay plugins... and hear the guitar thru all the above in realtime (as I'm playing)... without latency issues.

This is very flexible... as you can swap Cab IRs... and adjust Reverb/Delay at any point.  

The new AxeFX III lets you mix up to four simultaneous Cab IRs in each of two separate Cab Blocks.

The above method could be used to yield similar results.

With new capabilities come new opportunities...

Hi Clint,

The AudioBox VSL reports latency accurately.

4.9ms total round-trip latency at a 64-sample ASIO buffer size 44.1k

2 hours ago, Tezza said:

o be honest, I don't understand the obsession with software monitoring.  I must be missing something. Presonus takes the cake with it's Blue Z, Green Z and Native Z latency monitoring with it's drop out protection etc, then when you get problems you have to turn off plugins, freeze tracks and fiddle with dropout and buffers. Too complex and fiddly for me. 

With the "Hybrid Buffering" approach used by Studio One (Logic and Samplitude had this years ago, ProTools got it at version 11), you don't have the big CPU hit from monitoring thru software. 

• Tracks that are merely playing back are processed using a much larger buffer size.
• Tracks that are being monitored thru software are processed using the small buffer (only while being monitored).

This gives you the best of both worlds.

When you can monitor thru software (no glitches) at 1ms round-trip latency, it pretty much makes hardware based monitoring moot.

You can monitor thru AmpSim plugins in realtime... or anything type of processing... and never give it a second thought.

When you go to play a virtual-instrument... one-way Playback latency is incredibly low.

Some audio interfaces don't report their record latency accurately.

Check the Record Offset for your audio interface.

• Record a short high-transient spike signal (like an isolated click).
• Now, take that single "Click" and rerecord it (physically patch an output to an input) to a second track.
• Zoom way in... and measure the difference between the two clicks (in samples).

This is your audio interface's Record Offset.

In Sonar, under Preferences>Audio>Sync And Caching... enter the number of samples (Record Offset) in the "Manual Offset" box.

Newly recorded audio will now line up correctly.

The Roland audio interfaces have some nice features.

Lowest possible round-trip latency just isn't their forte'.

At a 48-sample ASIO buffer size 44.1k, round-trip latency is 7.4ms.

Rendering video makes even a large/dense audio project look light-weight.

IME, The ASIO implementation in Premier, After-Effects, etc is not quite on-par with better audio applications.

10 hours ago, Blades said:

In Cakewalk, the screen at 64 samples says "effective latency is 1.5ms".  It's not usable and I don't really understand how 2.9+2.5 somehow is 1.5, but whatever.  It's not real.

A 64-sample ASIO buffer at 44.1k = 1.5ms (This is true no matter what audio interface you're using)

While true (and I know you already know this), that's not telling the whole story.

When you're monitoring in realtime thru software EFX/processing, you're dealing with round-trip latency:

• ASIO input buffer (1.5ms)
• ASIO output buffer (1.5ms)
• A/D D/A (~1ms)
• Driver's safety-buffer - this is the X-Factor when it comes to round-trip latency and it's often hidden (can vary radically)

In this example, we're already at 4ms... without factoring in the safety-buffer.

If the audio interface is one of the better makes, the safety-buffer will be small and round-trip latency will be ~5-6ms.

If the safety-buffer is large, round-trip latency can be more than double.

2 hours ago, Tezza said:

For keyboards/drums, I use a nektar midi keyboard controller and Kontakt/Komplete and other VST's. Unfortunately, there is no way to avoid the latency with this setup. However, I have found the Steinberg drivers to be the best so far out of what I've tried on my system so I have no complaints. Don't really notice it, whereas i did notice it on the other audio interfaces i tried.

In the case of playing soft-synths, you're dealing with one-way (playback) latency.

Monitoring audio tracks in realtime thru software EFX/processing, you're dealing with full round-trip latency.

The UR44 (IIRC) yields about 7ms total round-trip latency at a 32-sample ASIO buffer size 44.1k.

Some of the absolute latest generation of AmpSim plugins are sounding pretty decent.

• Helix Native sound pretty good.
• The new version of TH-U (from Overloud) is going to have a feature similar to Kemper's ability to "profile" actual mic'd amps/cabs.
• PRS Super Models sounds good (IMO) if you use different Cab IRs

Onboard DSP to process/route/mix/loop-back-record can be extremely useful (if you use it).

If you're after lowest possible round-trip latency, onboard DSP will slightly increase it.

Part of the reason Quantum can achieve such low round-trip latency; it has zero onboard DSP for routing/mixing/loop-back-recording.

All monitoring has to be done via software.

Why so fixated on lowest possible round-trip latency?

In the case of Quantum, since all monitoring has to be done via software, it's critical.

Lets say you're using a Kemper Profiling amp... or something like Helix or HeadRush (all hardware guitar amp sims).

The Kemper itself can have up to 4ms round-trip latency.

The whole point of an audio interface like Quantum is to keep round-trip latency as low as possible.

If Quantum were yielding 4ms latency, add the Kemper's 4ms latency... and you're at 8ms total round-trip latency (while monitoring guitar).

That's a significant step backward compared to hardware based monitoring (8ms vs. near zero).

Since Quantum can actually get down to 1ms total round-trip latency, even with Kemper's worst case scenario, you're at 5ms total round-trip latency.

At 1ms total round-trip latency, Quantum makes software based monitoring effectively on-par with hardware.

Monitoring via software at 1ms round-trip latency hits the CPU hard.

High CPU clock-speed is critical... as this isn't a process well-suited for multi-threading.

In today's economic world, few have the luxury of being 100% altruistic. 

4 hours ago, Bill Ruys said:

I used to have a MOTU 2408 Mk3 and 24I/O connected via PCIe-424 audiowire.  Both would run at very low latency and if the product wasn't too loaded, I could run at 64 samples.  When I sold my studio, I purchased a 896Mk3 Hybrid, as I wanted some built-in preamps, plus I add more via ADAT when I need them.  I have never, ever been able to get anywhere near 4ms whether connected via USB or FireWire (TI Chipset).  I find I have to run with 256 samples at a minimum and often more to avoid dropouts or distortion.  I really miss the low latency I got with the audiowire gear.

However, if you're saying you can get sub 4ms with the MOTU USB stuff, maybe there is hope?

Only reason I have steered clear of the new MOTU interfaces is the crap-shoot that is thunderbolt on Windows.  Also, I currently run a Ryzen rig, which, of course, does not support thunderbolt.

The 896mk3 Hybrid interface (and all the Hybrid series) was (round-trip latency wise) a step backward from the original 896HD.

The original 896HD yielded 5ms round-trip latency at a 64-sample ASIO buffer size 44.1k

The Hybrid series added onboard DSP processing/mixing... and that increased round-trip latency to ~6.5ms at those same settings.

To get sub 4ms round-trip latency from MOTU USB, you have to be running one of the newer AVB models (or spin-offs).

The newer MOTU USB drivers allow you to tweak the safety-buffer size.

FWIW, Thunderbolt under Windows is not a crap-shoot. 

You just have to make sure you've covered all the details.

MOTU was one of the  first companies to have (release version - not beta) Thunderbolt drivers for Windows that support "PCIe via Thunderbolt" (allowing PCIe level performance).

16 minutes ago, Pathfinder said:

Echo Tech told me their card would NEVER work with a PCi to PCIe adapter. Uh, they were wrong.........

The reason they gave you that advice... 

Echo cards never worked with bridged PCI slots on motherboards.

Lynx cards also had major issues with bridged PCI slots.

RME and M-Audio were much more forgiving... and worked with most bridged PCI slots.

16 minutes ago, JCody said:

What's the best way to do this?  I've always started from a template with Kontakt loaded for several tracks. But when opening a midi file, nothing is loaded and I'd like to know the easiest way to assign each track to an instrument. I currently create a new track with the instrument and drag the midi clip from its original location. I'd rather find a way to assign the instrument to the original midi track.

• Add an instance of Kontakt (and load the desired sound/s)
• Add an Audio Track
• Assign the Audio Track to receive audio from that Kontakt instance (Input drop-down list)
• Assign the MIDI track to output to the instance of Kontakt (Output drop-down list)

This doesn't answer the OP's question... but he might want to check out Melodyne (light version included with CbB).

Melodyne is far more advanced (sounds better) for intonation adjustment.

Well placed overheads can capture the bulk of a good drum-kit sound.

This is a great example of that...

Well done

+1 scook

Open the type 1 SMF... and immediately save it as a Cakewalk Project.

Often times, there are unwanted Tempo, Patch Changes, and Volume/Pan MIDI events.

I'd strip those out unless specifically desired.

Then, assign the individual MIDI tracks to specific virtual or hardware instruments.

Delete CbB's "aud.ini" file.

You'll have to reset your audio preferences.

In similar cases, that often resolves the issue.

Hi Larry,

I get it...

When we play, we always hire commercial sound/lights.  Makes the whole process easier and more enjoyable.

Most of the local working bands here do the same.

When 1am hits, we (old folks) want to load our gear and head home.  

The band has to charge more, but that's offset by a competent engineer actually mixing the show (both audio and lights).

A great sound/light company helps fully maximize the band (which helps with draw/etc).

Was the original Standard MIDI File type 0 or type 1?

21 hours ago, Starship Krupa said:

I have 3 Windows systems, all were running Windows 7 until a month ago. Microsoft just did the same thing that BandLab does and gave me 3 free licenses for their current OS, Windows 10.

FWIW, I don't think the "Free" upgrade to Win10 was an altruistic move by Microsoft.

They're trying to get Win7,/Win8.1/Win10 users on a single platform.  Less to maintain/support 

FWIW, If the machine is having issues such as high DPC Latency, it doesn't matter what DAW software/plugins you're running (all with be negatively affected).

BTW, Unplugging the LAN port disconnects from the Internet... but it doesn't actually disable the LAN controller.

ie:  If the LAN controller's driver is causing high DPC Latency, simply unplugging the LAN port won't resolve it.

You can update/roll-back the LAN driver... or disable it (Device Manager or motherboard's BIOS) while working with audio.

On 2/3/2019 at 6:57 AM, Twisted Fingers said:

Jim this is very useful information for me. I have an ASUS X99 DELUXE II motherboard with the ASUS/INTEL ThunderboltEX 3 PCIe card. It's good to hear that, once the prices drop to within my price range, I'll be able to use a Thunderbolt interface and benefit from the extremely low latencies they are capable of.

That combination works great.  Used it myself for a good while...

On 2/2/2019 at 4:12 PM, Larry Jones said:

Eventually I figured out that if sound travels through normal air at normal temps at 1100 feet per second, that's roughly one foot per millisecond, so (for me) anything less than about 15ms is acceptable -- that is, I can easily play in time using amp sims, as 15ms is approximately the equivalent of standing 15 feet in front of your amp on stage, which I have done all my life. In some ways I envy people who can hear the difference a millisecond makes, but I can't, so my life is a lot easier.

If it doesn't bother you... it doesn't bother you

Most guitar players I've seen are either using wedge monitor or IEMs (in addition to their amp).

Our guitar player wants to hear the sound from his Amp (about 4' behind him)... plus in the wedge directly in front of him (doesn't use IEMs).

Monitoring via wedge or especially IEMs eliminates higher latency of standing 15' from an amp.

I like the tight timing from using IEMs, but I don't like the feeling of being isolated/separated from the audience.

On 2/2/2019 at 9:29 PM, chris.r said:

Beside hidden buffers there's one more factor in play, which is the on-board USB controller (I think it's called hardware controller). PCI card doesn't use any additional controllers while USB interface does have to go through it and your PC have to compensate that with it's CPU power, unfortunately. That's one more extra additional cost when comparing USB to PCI audio interfaces.

With any reasonably current (Intel) build, you'll have Intel USB-2 or USB-3 integrated into the motherboard's chipset.

That eliminates most compatibility problems.

Prior to Z77 chipset motherboards, literally all USB-3 ports were third-party add-on controllers... as USB-3 had not yet been integrated into Intel chipsets.

The CPU load from the USB controller is negligible on a reasonably current CPU.

If you run a dense audio stress-test using a PCIe card vs. RME USB-2 audio interface, you'll find you can run the same amount of processing with each.

Using the PCIe card doesn't result in more DSP processing power.

FWIW, The best USB-2 Audio interfaces *are* on par with the best PCI audio interfaces.

• M-Audio Delta and Audiophile series yields 5ms total round-trip latency at a 64-sample ASIO buffer size 44.1k
• RME USB-2 audio interfaces yield 4.9ms total round-trip latency at a 64-sample ASIO buffer size 44.1k

A 64-sample ASIO buffer is 1.5ms at 44.1k (doesn't matter that audio interface you're using).

Latency differences between various A/D D/A is negligible. 

When it comes to round-trip latency, the "X-Factor" is the driver's safety-buffer (which is often hidden).

The best audio interfaces can use a smaller safety-buffer.  Lesser audio interfaces use a larger safety-buffer (resulting in higher round-trip latency).

Round-trip latency is the sum of the following:

• ASIO input buffer
• ASIO output buffer
• A/D D/A converter
• The driver's (often hidden) safety-buffer

Do any of you remember the DAL CardD+?

The first "Pro" audio interface for the PC.  😉