USB Audio/ASIO in WinXP 32 Host? USB hardware gurus needed.
Posted: 27. Jul 2009, 08:42
Hi VBoxers!
This is my first post so go easy on me OK?
I've been trying to do some recording inside a Windows XP 32-bit Guest on both xubuntu 9.04 x64 and Linux Mint 7 x64 hosts. It hasn't worked out as planned and causes a bunch of glitches and dropouts in the audio stream. I've tried the Search and have come up dry. I think it would better to declare my desired end state before getting into the hardcore troubleshooting:
I want to record inside the Guest using a USB-based M-Audio i/o box so I can record scratch (throwaway) vocal and guitar tracks in my home studio over tracks I record elsewhere. I don't want the sound to come out of the host (I have an internal sound card that the host uses for system sounds) but, rather, directly out of the USB device used in the Guest.
Is there any way I can get direct access to the USB device without it being used by my host as well? Ultimately, I am looking for dedicated, direct hardware access to this device in the guest. I think the fact that the host is sending cycles to the hardware as well (even though it isn't being used by PulseAudio or ALSA) is where my latency issues are coming from. I essentially want to make the device invisible to the host because I need the latency on that device down under 20ms using ASIO drivers. Normal system sounds come through the device using DirectX or WDM drivers, but recording using these has a lag that renders it unusable. Using ASIO drivers on the guest causes major skipping in the audio stream and eventually causes my recording app (Steinberg Nuendo) to bomb out and stop recording.
For folks who don't record: If any of the ASIO/DirectX/WDM is confusing, think of it like OSS drivers versus ALSA drivers (Linux sound gurus, feel free to chime in here).
Any help you could shoot my way would be great. I've been trying various things on the host in the USB filters settings but have come up dry.
For info, I'm a bit of a Linux virgin (I've been using it full-time for about 6 months), but I'm not afraid of the command-line. In fact, I grew up using DOS and prefer the command-line to a lot of GUI-based configuration. Because of the workflow in our studio (the main engineer learned on Windows and intends to stay that way despite my FOSS/Linux diatribes), I need to have Windows-based tools available. The constant rebooting into a native Windows session is starting to wear on my nerves.
My machine is Phenom X4 940/MA770-based with 4GB RAM, SATA2 storage, NVIDIA 9600GT video and M-Audio sound (at home) and RME sound (in the studio).
Cheers,
--Sean
PS. Just a stream-of-consciousness moment for a sec: has anyone out there had any success in reliably devoting a single core (or pair of cores, for redundancy's sake) on a multi-core machine to system tasks while the other cores can focus solely on other tasks? Can the guest have sole access to the hardware without any overhead caused by the hypervisor?
This is my first post so go easy on me OK?
I've been trying to do some recording inside a Windows XP 32-bit Guest on both xubuntu 9.04 x64 and Linux Mint 7 x64 hosts. It hasn't worked out as planned and causes a bunch of glitches and dropouts in the audio stream. I've tried the Search and have come up dry. I think it would better to declare my desired end state before getting into the hardcore troubleshooting:
I want to record inside the Guest using a USB-based M-Audio i/o box so I can record scratch (throwaway) vocal and guitar tracks in my home studio over tracks I record elsewhere. I don't want the sound to come out of the host (I have an internal sound card that the host uses for system sounds) but, rather, directly out of the USB device used in the Guest.
Is there any way I can get direct access to the USB device without it being used by my host as well? Ultimately, I am looking for dedicated, direct hardware access to this device in the guest. I think the fact that the host is sending cycles to the hardware as well (even though it isn't being used by PulseAudio or ALSA) is where my latency issues are coming from. I essentially want to make the device invisible to the host because I need the latency on that device down under 20ms using ASIO drivers. Normal system sounds come through the device using DirectX or WDM drivers, but recording using these has a lag that renders it unusable. Using ASIO drivers on the guest causes major skipping in the audio stream and eventually causes my recording app (Steinberg Nuendo) to bomb out and stop recording.
For folks who don't record: If any of the ASIO/DirectX/WDM is confusing, think of it like OSS drivers versus ALSA drivers (Linux sound gurus, feel free to chime in here).
Any help you could shoot my way would be great. I've been trying various things on the host in the USB filters settings but have come up dry.
For info, I'm a bit of a Linux virgin (I've been using it full-time for about 6 months), but I'm not afraid of the command-line. In fact, I grew up using DOS and prefer the command-line to a lot of GUI-based configuration. Because of the workflow in our studio (the main engineer learned on Windows and intends to stay that way despite my FOSS/Linux diatribes), I need to have Windows-based tools available. The constant rebooting into a native Windows session is starting to wear on my nerves.
My machine is Phenom X4 940/MA770-based with 4GB RAM, SATA2 storage, NVIDIA 9600GT video and M-Audio sound (at home) and RME sound (in the studio).
Cheers,
--Sean
PS. Just a stream-of-consciousness moment for a sec: has anyone out there had any success in reliably devoting a single core (or pair of cores, for redundancy's sake) on a multi-core machine to system tasks while the other cores can focus solely on other tasks? Can the guest have sole access to the hardware without any overhead caused by the hypervisor?