Stable low latency in Win 10

[ejinbc]

I wanted to share some of my recent explorations towards achieving lower latency while recording a band in PT 12.8 / Win 10 using a mix of VIs and 12 x 24/96 kHz audio channels. We monitor drums (TD6V), keys (2 players), acoustic guitar, and 4 x Vox live through PT during recording. Please consult my sig for setup details.

With a MOTU 1248 (latest drivers and up-to-date Win 10) I can record without clicks or pops at 24/96 kHz with a 64 sample audio buffer and 32 sample PT output buffer (MOTU allows these to be set separately) for at least 3 hours continuous without issue. Some things that have helped:

1) Use the internal mixer in multitimbral instruments – do not send the individual channels from instruments into individual Aux channels in PT during low latency recording. For example, do not send the snare, kick, toms, etc, out as separate channels to be processed in PT separately during recording. Only return a stereo pair from Kontakt (i.e. do the drum mix in the Abbey Road mixer). This seemed to help with CPU load. We use 2 separate instances of Kontakt on Aux channels when recording (Drums and Keys).

2) Use a CPU affinity of ‘FC’ when launching PT – this frees up one core for background audio processing on an i7 CPU. This is required on my system (see sig) to achieve click/pop free operation at low latency – otherwise I get random CPU spikes which will halt recording.

3) Minimize the use of subs and sends. Recently on GS someone was discussing optimizing your DAW setup to minimize serial processing of audio signals – the more parallel the signal path, the easier it is for the program to schedule across all cores (makes sense). So I route all signals straight to Master Output channels for monitoring. I have one send setup for Vox slapback (HDelay) echo during recording for monitoring.

4) Read the chapter on Groups in PT. Groups can be used for level control to minimize the need for Aux channels – left click to adjust the group (like a VCA), and right click to adjust an individual group members level (wow). I have also started to mix more like this and for many tasks it keeps the mix simpler and easier for me to control.

Hope this helps someone.

E

>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> >
Machine Details: Asus P8Z77 MB, 3770K CPU O/C 4.5 Ghz, 16 GB RAM, 512 GB SSD, 2 TB Data drive, Gigabyte 950 video card running HDMI 2.0 @ 4K resolution, all USB peripherals plugged into USB3 ports EXCEPT Motu 1248 which is plugged into rear USB2 port, Turbo Mode Off in BIOS, C-states Off in BIOS, WiFi Off in BIOS, BT Off in BIOS, On-board sound Off in BIOS, HPET Off in BIOS. Maintained offline only.

Midi input controllers: Command 8, Roland TD6V, Edirol PCR M80, Korg SP250, Korg Nano

Software: Windows 10 (all standard DAW optimizations), Pro Tools 12.8, Waves Gold Bundle + Waves HEQ, Native Instruments Komplete 10, …

Session details for typical low latency recording (64/32 buffer): 24 bit @ 96 kHz, 12 Audio Tracks, 3 Aux VI Tracks, 16 MIDI Tracks, 6 HEQ and 1 HDelay instance.

[YYR123]

Good deal, so you are pulling your headphone mixes from the MOTU mixer or your PT Mixer to MOTU outputs ?

[ejinbc]

YY

I take everything from the PT mixer routed directly to different outputs on the MOTU. These sessions are 'live recordings' so I don't set up custom headphone mixes. I use three outputs from the MOTU to feed monitors - Drums (mono), Keys and Vocals (stereo); panning and levels are adjusted by the musicians during the session using a couple wireless mice and a Command |8. Everyone can see the mixer on a 55" 4K display in the studio.

I had used the internal MOTU mixer (zeroish latency) for a while for monitoring vocals, but I find it much more convenient to keep everything together in PT during recording.

I posted to give some ideas to people trying to use vanilla PT with VIs (and vocal monitoring) with minimum latency. For my system the processor affinity setting and keeping the routing as simple as possible in PT while recording have helped.

Cheers