Jump to content
  • 0

Cakewalk Latency recording from Roland TD-27 Multi-Track Audio


Alex Cordero

Question

MY GOAL:
Record *Multi-Track Audio* from my TD-27 Drumkit so I can mix and add VSTs as needed.
I've been recording L/R 1/4" out from the TD-27 all this time but the mix is not the best and I'm limited to what I can do with it on a single stereo track.

EQUIPMENT:
Roland TD-27, Good quality USB Cable, Brand New Gaming PC with 64Gb Ram, AMD Ryzen 9 9950 on a Asus X870 Gaming MB & Win11 Pro. (No Audio Card - just onboard), Focusrite Gen 3 2i2, and of course, Cakewalk.

SETUP:
The TD-27 is setup for Vendor Mode (which they suggested) and has updated firmware. Connected via USB directly to PC. The PC connects to my Focusrite which outputs to speakers. I've downloaded the latest TD27 drivers and rebooted/restarted everything. Windows is also up to date with no driver issues. Cakewalk is also up-to-date.

THE ISSUE:
When I go into Cakewalk preferences and select ASIO, I can play back all my other instrument tracks (bass, keys, guitar, vocals, etc), *BUT I can't select any individual drum tracks as they're greyed-out.
Selecting WASAPI Shared, I now can select the individual drum audio tracks, *BUT I get terrible *LATENCY* and can not adjust the buffer as it's greyed-out.
WASAPI Exclusive also allows me to select the drum audio tracks, *BUT now sounds horribly broken up and still has latency. Although I can adjust the buffer, it still sounds absolutely unusable.
I've downloaded ASIO4all, *BUT it didn't work at all. 

I've also tried on another computer (My new computers predecessor which has all the same software and setup, *BUT same result.
I can record no problem from L/Mono, Right, But that isn't what I'm trying to do.
I can also get Midi to record into the track, But also, not what I'm trying to do.
I've search forums and YouTube and can't find the answer.

Any suggestions are appreciated. Thanks

Link to comment
Share on other sites

Recommended Posts

  • 0

The drastic difference between 2x the buffer latency (2 x 5.8ms = 11.6ms) and the reported round-trip of 28.1ms indicates that the TD-27 has a lot of hardware/firmware latency. Typical USB interfaces like your Focusrite will add only 6-8ms to the buffer latency, and good ones may add only 3-4. Adding 16.5ms is huge, and you're not going to eliminate that or compensate by running a smaller buffer (256 samples is a lot for real-time monitoring) . I can only think the TD-27's audio output is really only meant for recording-pre-existing MIDI and not real-time monitoring.

I would suggest your use the Focusrite to record MIDI in real time, monitoring the combined stereo analog output from the TD-27 through Sonar (or direct from the Focusrite with direct-monitoring engaged) and after polishing up the MIDI, switch to the TD-27 and record it's separate outputs. I realize this is awkward and not what you would like, but based on what the Driver Setting page is showing, you are not going to get real-tme playability with the TD-27 driver.

You could try recording the multitrack audio from the TD-27 with its driver selected while direct-monitoring from it's headphone out. The recordings will be automatically compensated for Input latency when you stop recording, but you will probably find that the MIDI is late because you're performing to a delayed late click from Sonar due to the Output latency and MIDI is not latency compensated. If you're not planning on editing the MIDI and re-recording maybe that doesn't matter to you.

Edited by David Baay
  • Like 1
Link to comment
Share on other sites

  • 0
9 minutes ago, David Baay said:

I would suggest your use the Focusrite to record MIDI in real time, monitoring the combined stereo analog output from the TD-27 through Sonar (or direct from the Focusrite with direct-monitoring engaged) and after polishing up the MIDI, switch to the TD-27 and record it's separate outputs. I realize this is awkward and not what you would like, but based on what the Driver Setting page is showing, you are not going to get real-tme playability with the TD-27 driver.

This is probably your best bet if that driver isn't up to snuff.  But...

Out of curiosity....does this update apply to your TD-27...and have you installed it yet?  Sorry.  Couldn't post a link.

image.thumb.png.f71d77c3dfe4e8e5fde1434c7ce28d7f.png

 

 

Link to comment
Share on other sites

  • 0

with drums, even a 2ms latency will be audible to most people. so a 28+ms latency will be nearly impossible to ignore (basically the premise for this entire thread).

and as an ahole, i'll repeat my suggestion of monitoring live the drum performance vs the other instrumentation, and if need be, include the metronome in the playback. this way you're playing to the material (without looping backing) and hearing your drumming live via the analog path in your IO. turn off DAW monitoring of your MIDI and drum audio from the DAW. 

  • Like 3
  • Great Idea 1
Link to comment
Share on other sites

  • 0

Agreed completely Glen.  

The latency does not matter unless you engage INPUT ECHO which is a bad idea.  You always monitor audio recording at it's source, not after it has traveled through 6 different pieces of technology that each  will add latency.  And ya, those RTL specs are abnormally high but my guess Roland hasn't updated how they make their driver since they owned Cakewalk.  

I personally don't see the OP having any issues recording if they use the TD27 headphones to monitor the kit and Cakewalks playback. I'll almost guarantee the Midi and audio tracks will be super close to perfectly synced to the project. 

Pretty easy to test. 

And this is why everyone who uses digital kits will always recommend not using a VST to create the sound while recording. It will be horribly late. Unless you have a million dollar audio system and get only 3ms of RTL without dropouts. The average interface these days can get close to 6ms RTL but most of us are lucky to get 8, or even 10 without drop outs in a half finished project. That is certainly enough to throw off most musicians timing.  

Edited by Sock Monkey
  • Like 1
Link to comment
Share on other sites

  • 0
5 hours ago, Sock Monkey said:

Sounds like you only missed the small detail of clicking "Apply" 

It all looks good to me.

Few things I would add.  In the Sync and Caching. Make sure the TD 27 shows as Latency Adjustment device. This is critical for being in sync.  

Screenshot(51).png.efb207f896d55bbecc28db5fd45a9cfb.png

Also if you want to record and playback midi make sure that the TD27 is selected as an input and output device. Always hit Apply after making changes. 

Screenshot(52).png.d2372004f00d2da7df7781f666262009.png

Awesome... didn't even see the sync/cache setting there. It shows both the FR and TD as options.
The Midi part is selected.

Thanks!

  • Like 2
Link to comment
Share on other sites

  • 0
5 hours ago, David Baay said:

The drastic difference between 2x the buffer latency (2 x 5.8ms = 11.6ms) and the reported round-trip of 28.1ms indicates that the TD-27 has a lot of hardware/firmware latency. Typical USB interfaces like your Focusrite will add only 6-8ms to the buffer latency, and good ones may add only 3-4. Adding 16.5ms is huge, and you're not going to eliminate that or compensate by running a smaller buffer (256 samples is a lot for real-time monitoring) . I can only think the TD-27's audio output is really only meant for recording-pre-existing MIDI and not real-time monitoring.

I would suggest your use the Focusrite to record MIDI in real time, monitoring the combined stereo analog output from the TD-27 through Sonar (or direct from the Focusrite with direct-monitoring engaged) and after polishing up the MIDI, switch to the TD-27 and record it's separate outputs. I realize this is awkward and not what you would like, but based on what the Driver Setting page is showing, you are not going to get real-tme playability with the TD-27 driver.

You could try recording the multitrack audio from the TD-27 with its driver selected while direct-monitoring from it's headphone out. The recordings will be automatically compensated for Input latency when you stop recording, but you will probably find that the MIDI is late because you're performing to a delayed late click from Sonar due to the Output latency and MIDI is not latency compensated. If you're not planning on editing the MIDI and re-recording maybe that doesn't matter to you.

I definitely plan on editing and quantizing.

Thanks for the feedback.

Link to comment
Share on other sites

  • 0
5 hours ago, HOOK said:

This is probably your best bet if that driver isn't up to snuff.  But...

Out of curiosity....does this update apply to your TD-27...and have you installed it yet?  Sorry.  Couldn't post a link.

image.thumb.png.f71d77c3dfe4e8e5fde1434c7ce28d7f.png

 

 

Thanks Hook!

I do have V2 installed.

Link to comment
Share on other sites

  • 0
5 hours ago, Glenn Stanton said:

with drums, even a 2ms latency will be audible to most people. so a 28+ms latency will be nearly impossible to ignore (basically the premise for this entire thread).

and as an ahole, i'll repeat my suggestion of monitoring live the drum performance vs the other instrumentation, and if need be, include the metronome in the playback. this way you're playing to the material (without looping backing) and hearing your drumming live via the analog path in your IO. turn off DAW monitoring of your MIDI and drum audio from the DAW. 

I did mess with the monitoring on both the FR and in CW. Maybe I had missed something there in my frustration.

Link to comment
Share on other sites

  • 0

I'm going to guess this was just a weird gremlin. At this point I can't even force back the latency issue. (Hopefully it never comes back ;) )

Maybe you guys are good luck.

Thanks everyone for the great input! Still maybe this post will help someone else.

 

  • Like 1
Link to comment
Share on other sites

  • 0
17 hours ago, Alex Cordero said:

So, a few hours ago the latency stopped. No other updates, changes or anything I haven't tried before... except having L/R and USB connected at the same time. Maybe has nothing to do with it.

I have it currently recording Audio (not Midi) to multi tracks. (I'm ok with that for now but Midi would be the next step)
The TD27 simultaneously has L/R connected to the Focusrite and USB connected to the PC. Both options work. (Just seeing what worked and didn't)
It's recording from the TD27 into Cakewalk while the previously recorded tracks AND the sound from the TD27 is routed through the Focusrite coming out of the speakers.

I missed this post. Sounds like you are now direct-monitoring the TD27's analog output through the Focusrite which doesn't require Sonar to be using its driver. If you enable Input Echo on the tracks receiving the multitrack input from the TD-27's driver, you'll hear an echo/delay from the latency. Leave Input Echo off, and it's a perfectly valid solution apart from not being able to record anything else from the Focusrite's inputs, not being able to have FX applied to the drums in real time, and having some additional output latency from soft synths if you ever play/record live from a MIDI keyboard.

  • Like 1
Link to comment
Share on other sites

  • 0
5 hours ago, Alex Cordero said:

I did mess with the monitoring on both the FR and in CW. Maybe I had missed something there in my frustration.

You want to direct monitor as close to the source as possible (i.e., the headphones out on the TD-27 itself when tracking), and disable input echo (and even mute) the tracks in the DAW. Stupid side note here but need to state it... there is sometimes a vast difference between tracking and post-production (just listen to some raw stems of production work). Even simple FX can add latency... the point of tracking is not to hear the "final product" but to capture the performance (just need to capture the (raw) audio from the kit that will feed post-production).

  • Like 1
Link to comment
Share on other sites

  • 0

Since the path from TD-27 to and through the Focusrite's direct monitoring circuit is all analog, monitoring from the Focusrite will be the same as from the TD-27 in terms of latency and will be necessary to hear Sonar's click or existing tracks. I agree tracking is often best done "dry" but there are situations where you really need to hear FX (e.g. syncing the performance tempo to a delay when not playing to a click), and it can help the performance to  be hearing some semblance of the reverb space that will be on the final mix as it can affect how you articulate. You just need to stay away from FX that use big lookahead buffers.

Edited by David Baay
  • Like 1
Link to comment
Share on other sites

  • 0

For "Sync and Caching" / "Record latency"  (appearance can be confusing):

Roland is not known for low latency, but as suggested before, you can try to tune the buffer size. Use Rolands own ASIO (sometimes can be opened from Cakewalk, sometimes from system tray only, note you need to "Stop audio engine" in Cakewalk during tuning, there is an icon in the Transport module). How far you can get depends from your computer audio optimization (default BIOS/Windows settings are not audio optimized).

Audio interface latency is important only for "throw the DAW" monitoring during recording. Since you record TD sound, you probably want that sound... and so you probably want to listen it during recording. And so there is no reason to route it throw the DAW (during recording). Headphone/speakers directly on TD or mixer (FR can play that role) are not influenced by interface latency.

Played sounds (f.e. metronome, previously recorded tracks, etc.) and recorded material will be "at correct time" as long as interface latency is known to Cakewalk, it will compensate (ASIO reported latency is more or less correct for TD and FR). You can also mix some "not time critical" effects (like reverb) from live (drum) sound, just don't use in DAW dry sound during recording.

If you want record MIDI, record it while still monitoring original TD sound. For monitoring with drum VSTi  you will have to tune latency. MIDI latency is in general "gray zone", it is not reported and unclear how it is accounted. And it can be significant. Later quantized by audio buffer size, delayed by audio output latency, apart from possible "plug-in delays" (keyword PDC), the result can be disappointing. I mean that is not worse the effort in case of "self sounding" source recording.

 

On 1/4/2025 at 7:43 PM, Glenn Stanton said:

with drums, even a 2ms latency will be audible to most people. so a 28+ms latency will be nearly impossible to ignore (basically the premise for this entire thread).

Note that the first sentence is "most people may notice extra 2ms latency under some conditions". "Natural latency" of acoustic drum set is around 1-2ms between peaces plus 1-2ms for all (sound speed is ~34cm/ms). So in practice if your e-drums have 2-4ms latency and you play with headphones, the latency will be not worse then "natural". When you change from headphones to monitors, the latency will increase by yet another 2-4ms (the distance from your head to monitors). Yet most people still do not perceive that as "disturbing". But there will be person dependent "border" after which you notice "unusual delay". And when extra 2ms cross that border, you notice that "extra 2ms".

Our brain is "smart compensating" "natural" latency of the sound you listen (just imagine you can really feel 1ms latency... you will be unable comfortably  listen someone who is moving more then 30cm when speaking/singing without mic). So critical is your own voice only - the distance is small and fixed (head size) and the brain knows it ;) 

  • Like 1
Link to comment
Share on other sites

  • 0

Natural latency was sort of never really questioned by most musicians. Guitar players and worse, bass players if given a big enough stage might stand 15 or 20 feet away from amps. Just think of the days before everything was put in the PA and then floor monitors? 
 

I realize later why as a bass player I felt more in the pocket if I stood close to my amp and the drums. 
 

When I got my digital drums I learned fairly quickly that you had to play them using the built in sound. 
Playing a VST even at the lowest buffer was just not in the pocket and you quickly loose confidence in your performance. 
Worse was when I tried to listen to both together and the slap back echo.  
 

In ear monitors are the way to go and especially if you’re using an analog mixer. A lot of mixers are digital now and when I use them I hear that same slap back due to the conversion and processing. 
 

Edited by Sock Monkey
  • Like 1
Link to comment
Share on other sites

Please sign in to comment

You will be able to leave a comment after signing in



Sign In Now
×
×
  • Create New...