Jim Roseberry

If your audio interface has a control panel with input channel meters:

Play the keyboard and look at those input meters.

If you only see a single channel via the input meter, the issue is keyboard/cable (outside of Cakewalk).

If Cakewalk is recording two channels on said track... and one is as-expected  and the other silence (noise-floor), the issue is most likely outside of Cakewalk (meaning the keyboard/cable/etc).

Recording the Left and Right outputs from a synth/etc to separate tracks allows a bit more control (without using additional plugins like Channel-Tools).

• You can pan hard left/right... or collapse the stereo image
• You can have separate processing on the left and right channels
• You can control the level of the left and right channels (independently)

There's no right/wrong... just a matter of which is more convenient.

12 hours ago, Colin Nicholls said:

I would also need a good PCIe thunderbolt card.

Quantum is a great choice... as long as you understand the one weak point (there's no onboard DSP for routing/mixing/loopback-recording).

ALL monitoring has to be done via software.

IOW, If you have an Axe-FX or Helix (guitar processor)... or a keyboard... and just want to sit/play (monitoring thru the Quantum), you'll have to fire up your DAW and use it to setup the desired monitoring (Levels/Routing/etc).

Note:  You can't add a PCIe Thunderbolt controller to just any motherboard.

• Motherboard has to specifically support a Thunderbolt-3 controller
• Motherboard must have a Thunderbolt-3 header that matches the type used by the Thunderbolt-3 add-in-card

The Thunderbolt-3 add-in-card resides in a full-length PCIe slot... AND has to be connected to this Thunderbolt-3 header on the motherboard.

Thunderbolt-3 works great on a Win10 PC. 

Just make sure you've covered all the details.

Just a couple of comments:

Video capture software puts substantial load on a machine.

Working with HD/4k Video is much heavier load than a typical audio project.

Running both a DAW project and video capture simultaneously will push some machines to their limit.

Increasing the ASIO buffer size allows your machine to better mitigate high CPU load.

Using an extreme example:

• Audio interface set to 64-sample ASIO buffer size
• Sample-rate 44.1k

When running DAW software, your machine has 1.5ms to process the next audio buffer and get it cued for playback.

If anything interrupts this process, you'll hear a glitch or (worse) experience a drop-out.

The lower the latency, the less efficient the load can be spread across multiple cores.

Thus, when working at lowest latency settings, CPU clock-speed is the single most important factor.

Can't go wrong with K240s.

I've used them for decades... but tend to use them for another listening perspective (not to actually mix).

When mixing, make sure to vary the volume at which you monitor.

ie:  Turn the level significantly down.  Can you still hear all instruments.

Over time, our ears become less sensitive... so there's a tendency to keep increasing the monitor level (same as playing live gigs with amps).

You can help save your ears (and mix) by turning the level down... and making sure the drums/bass/etc don't disappear.

On 8/26/2020 at 2:56 PM, micv said:

Would be great if the track's FX bin can be configured to be "post send" so the raw track can have its FX also.

If you have a Send set to 0dB (no gain change), what arrives at the Return (Stereo Bus) is a "mult" or copy.

You've got processing on the original track (which happens before the Send).

There's no way to make the original track's EFX Inserts "post send"... as the (Send) signal has left that channel.

On the Return (Stereo Bus), you've got EFX Inserts... and you can Send to another Return (Stereo Bus).

Let's say you've got a dry DI electric bass track... but want it to sound a bit more like Chris Squire or Geddy Lee.

You could create a Send on the DI electric-bass track... to a Return (Stereo Bus) called "Dirt".

On the "Dirt" return, use the EFX insert to add your favorite distortion plugin.

Adjust the level of the "Dirt" Return... to mix in the desired amount of distortion (added to the original  DI bass).

Upon listening to the distortion, it's affecting the bottom-end too much (we want to limit the distortion to effecting only the mids/top end).

In the "Dirt" Return's EFX insert, add your favorite EQ before the distortion plugin... and use a high-pass filter to roll off everything below 1k.

Now, the distortion is adding character... without losing clarity on the bottom-end.

On the Bridge of the song, the bass is playing a melodic part... so we want to add some Chorus (just to the distorted mult).

Create a Send from the "Dirt" Return... to a new "Chorus" Return.  Insert your favorite Chorus plugin in the EFX insert of the "Chorus" Return.

Adjust the level of the "Chorus" Return to balance with the original bass track.

This is what we've created:

• Original DI electric bass track
• Mult (copy) that's 100% distorted
• Chorus that's applied only to the distorted Mult

You may know all this...

I just wanted to use an example to explain what's possible. 

I'd be surprised if you couldn't accomplish what you're looking for... in a multitude of ways.

Worst possible scenario, you can create multiple physical copies of the original track (I doubt this is necessary).

Don't know of a way to prevent the issue... but it's  easy to grab the end of the resultant MIDI clip (with Snap to bar enabled) and drag to the desired bar.

The bounced clip ends at that point... because there's no other MIDI data beyond.

16 hours ago, kzmaier said:

Behringer U-Phoria UMC1820 - Good price and reviews!  Are drivers up to date for Windows?

PreSonus Studio 1810c - Like the input meters!  Would be happy with 1810 but cannot find one.

Focusrite Scarlett 18i8 3rd Gen - Had good luck with 6i6!

The U-Phoria and Scarlett will have about the same round-trip latency ~6ms (when set to minimum buffer size at 44.1k).

Presonus Studio 1810c round-trip latency will be a couple milliseconds higher.

Fidelity wise, it's a wash.

Driver wise, it's a wash.  None have the pedigree of RME.

Out of those choices, I'd go with the Behringer U-Phoria.

If you've used any of the X-series digital consoles playing live, you've used some of the same technology (mic preamps, etc).

2 hours ago, Bapu said:

My lovely lady did purchase a few of my basses (and one guitar) for me as gifts. I just keep most of the  others hidden.

I have a good friend that wanted to have guitars shipped to my address.

I let him do that a couple of times... then put an end to it.

I told him his wife is intelligent enough to know that his guitars don't reproduce.

She's an accountant... and probably already had a spread-sheet with all accounted.   😄

Not a VI, but the Behringer Poly D isn't terribly expensive... and it's an absolute blast to tweak and play.

Most fun I've had with a synth in a long time...

As has been mentioned, use Task Manager to have a look at the amount of RAM used by your largest projects.

You need enough RAM to avoid hitting the VM Swapfile (in lieu of enough physical RAM)... as that'll kill performance.

If your largest project uses 12GB RAM (and you've currently got 32GB)... bumping up to 64GB will have zero effect on performance.

I've had a MODX and current have a Montage (same basic hardware).

Both work fine with Cakewalk by BandLab.

The MODX USB connection can function as both an Audio Interface and MIDI I/O.

Many folks just use the USB connection for MIDI I/O... and use a separate (dedicated) Audio Interface.

On the MODX, go to Utilities>MIDI I/O, make sure MIDI In/Out is set to USB.

Connect the MODX via USB and load the driver (download the latest version from Yamaha).

Look in Device Manager (under Sound Video And Game Controllers)... and make sure the MODX is listed with no yellow exclamation points.

If you don't see the MODX listed in Control Panel>Device Manager, there's something wrong with the USB connection.   If it's not listed, as far as the machine is concerned, it doesn't exist.

If the MODX is listed in Device Manager, it's installed/working.

In Cakewalk>Preferences>MIDI Devices, make sure the MODX is enabled as both MIDI input and output devices.

Open a new project in Cakewalk.

Add an Instrument Track

On this Instrument Track, click on (enable) the Input Echo button.

Set the Instrument Track's MIDI input to be MODX>MIDI Omni

If you now play the MODX, you should see the Instrument Track's LED peak-meter showing activity (MIDI data is flowing to that track).

If you were running your audio interface at much higher buffer size... and/or using latent plugins in the project (especially in series), the latency can become much higher (and could get to the point were it sounds unnatural).

That's why I brought this up.

If you understand the concepts behind it, you can avoid (latency related) pitfalls.   

10 minutes ago, sadicus said:

So the idea is that the Latency still exists but the reverb blends it together so it's not noticeable?
Pre-Delay is something I just learned about and is so important for Recording Classical guitar, getting the transients.

The OP's video shows how you can combine two sources of monitoring

• Direct from the audio interface - dry signal (near zero latency)
• Signal processed thru DAW (in this case with reverb set 100% wet)

You need the DAW processed signal to be 100% wet (no dry signal).

If the reverb contained any dry signal, it would cause comb-filtering (unwanted phasing/chorusing).

• The signal direct from the audio interface is near zero latency.
• The signal processed thru the DAW is subject to ~5ms round-trip latency.

Had the Reverb contained any dry signal, it would be mixing dry vocal back in... but delayed by ~5ms.

By keeping the Reverb signal 100% wet, only the reverb is subject to the ~5ms round-trip latency.

• Dry vocal = near zero latency
• Reverb = ~5ms latency

In real physical spaces it often takes a few ms for the reverb (ambience) to reach your ears.

Thus, that ~5ms latency (in this example) wouldn't sound unnatural. 

The OP is combining hardware-based (dry signal) and software-based (100% wet signal) monitoring.

To clarify, the 100% wet signal *is* subject to round-trip latency... but in the case of Reverb, you probably won't notice a few extra milliseconds of "pre delay".

ie:  At 44.1k using a 64-sample ASIO buffer size, round-trip latency for many audio interfaces is ~5ms.

The 100% wet Reverb signal is subject to that ~5ms latency... but (again) you most likely won't notice it... as it sounds like you dialed in an additional 5ms of "pre-delay".

For those not familiar, Pre-Delay is a preset amount of time... before the reverb decay happens.

Adding some pre-delay allows transients to come thru clean/clear... as they're not immediately masked by the reverb.

1 hour ago, Cristian said:

Also LOL the thought of even trying to compare the i9-10900 with these embedded CPUS 😂

Yeah, just trying to make the point as to why I'm not super excited by Machine +.  😉

It'll be enough to wet-the-appetite... but not capable of anything remotely close to a full-fledged DAW.

Lest I sound anti-Machine, I do like Machine as a "finger-pad" controller.

Best feeling pads of anything currently available...

There's very little performance difference between DDR4/2400 vs. DDR4/3200.

If you're looking for rock-solid stability, NEVER over-clock RAM.

34 minutes ago, Cristian said:

I'm not really complaining. To be fair to NI, the price is competitive to MPC Live and MPC X.

Machine + will be able to run decent (not amazing) Virtual-Instruments and EFX.

MPC Live has (at least when I owned it) no ability to run Virtual-Instruments... and the onboard EFX/processing was basic.

• Record/Playback 8 tracks of Audio
• Trigger Samples

IMO, Neither MPC Live nor Machine + have enough processing power to get excited about.

MPC Live could be flaky.   Sometimes, it would power-up and there'd be no sound (have to reboot to regain playback).

34 minutes ago, Cristian said:

TBH my only question would be why NI went with the Z8350 and not the N4000 as a CPU, which is newer (2017 vs 2015) and better with similar costs.

N4000:

• Clock-speed = 1.1GHz
• Max Turbo = 2.6GHz
• 2 cores
• 2 processing threads
• Max RAM = 8GB

Z8350:

• Clock-speed = 1.44GHz
• Max Turbo = 1.92GHz
• 4 cores
• 4 processing threads
• Max RAM = 2GB

With such low clock-speed, neither CPU is well-suited for working with low-latency audio (small buffer sizes).

At larger buffer sizes, the two additional cores on the Z8350 would allow greater loads.

N4000 has slightly higher Turbo (Boost) frequency... but lower Base clock-speed.

Tight enclosure means relatively small cooling; neither CPU will run Max Turbo for extended periods.

As a point of reference, the new i9-10900k will run all 10 cores (20 processing threads) locked at 5.3GHz.

When working at smallest ASIO buffer sizes, clock-speed is the single most important factor.

ie:  Working at 96k using a 32-sample ASIO buffer size isn't something that lends itself to being heavily multi-threaded (spread across cores).

In this scenario, the 32-sample buffer size means the CPU has 1/3 of a millisecond to process the next audio buffer and get it cued for playback.

If anything interrupts this process, you'll experience a glitch.

More cores is beneficial, but not at the expense of significant clock-speed.

Performance increase from adding cores doesn't scale 1:1.  IOW, Doubling the number of cores doesn't double performance.

Generally speaking, the more cores... the harder it is to achieve highest clock-speed (especially across all cores).

In a perfect scenario, you want highest clock-speed... and most cores available.

25 minutes ago, craigb said:

Heh, yeah...  Over $9k for a Neal Schon model! 

I'm not usually into signature instruments.  No way I'd spend $9k for a Les Paul.  Well... maybe a real '59 (worth well into 6 figures).  😂

When I got that R9, it was on deep discount at GC.

Had been sitting in the Platinum Room for a long while... and had a couple of tiny dings.

Didn't pay anywhere near the current $6500 price.

Had an Alex Lifeson LP for a good while.

It was an exceptional guitar in most ways.

One thing drove me crazy, it picked up router noise (poorly shielded).

Best sounding Piezo of any instrument I've owned.

My router is in the studio room... and it was nothing but that (rhythmic) "tick, tick, tick, tick...."

Haven't checked out the 2020 LP Standards.

Had a 2014 R9 a while back.

IMO, That R9 was the quintessential classic-rock guitar. 

After owning/playing it, Standards (as gross as this sounds) felt/sounded "cheap" (by comparison).

Sold the R9 as I prefer the ergonomics of PRS.

The 2020 LP Standards look nice.

I hope Gibson is going back to the fundamentals... and focusing on quality.

When I was actively looking, Gibson Custom was just on a different level.

Given the cost, I guess you could (should) expect that...

I'm going to buck the trend.   😉

Why reinvent the wheel?

You've got your choice of many different 3rd-party "Samplers" (virtual instruments) that are FAR more advanced/evolved than a rev. 1 release.

FWIW, I don't want to be tied to a proprietary sampler, with limited function, that only works with one host software.

To me, that's taking a step two decades backward.

Development hours are somewhat a "precious commodity".

I'd rather the bakers focus on Cakewalk (DAW) which is their forte'.  

Let other companies (who specialize in Vi's)... do what they do best.

A cardiologist can treat you for the Flu.

Is that really the best use of his/her time???    😁

Samsung "Pro" series will not offer any speed improvement vs. the EVO.

The Pro series has double the warranty (10 years).

2 hours ago, Tony Carpenter said:

I have the 10940x 14 cores, it's a BEAST :).

Agreed.