Hi Guys.
I work for an audio company and currently doing a wealth of compliance checks on a new USB DAC product that we will be soon releasing. The DAC is capable of streaming 352.8 and 384kHz audio natively.
We have observed two issues with the WASAPI 3.1 plugin and using WASAPI Event mode:
a. Intermittent dropouts at high sample rates (typically 192K and above on some systems. The setting in Preferences->Advanced->Playback->Wasapi->hardware buffer in ms (event) is set by default to 10ms and seems to be not helping the issue. If set to 25ms then all the test machines I have will stream audio for day and pass bitperfect integrity tests. Laptops with dynamic power management seem most affected. To give a more plug and play solution a default that worked consistently on a wider range of machines would be appealing.
b. Intermittent dropouts when streaming 16bit 352.8 and 16bit 384kHz files with WASAPI plugin configured for 24bit output. If 24bit files of the fore-mentioned are played then all is fine, it's only when doing this 16bit to 24bit padding does it get unreliable on some machines. This indicates to me the code doing this conversion needs optimising or has a very small window of time to do the conversion. At lower sample rates (192K and below) it works ok.
As a reference we have also tested using J.River 18 using the same tests and the above seems reliable, We have also been pushing the driver through the Microsoft HCK tests to ensure we have good compliance.
Philip, the guy doing the driver has already posted on this forum reporting the bug we found in v3.0 plugin of it stalling intermittently changing sample rates. Thxs for fixing this and up issuing to v3.1.
Best regards
Steve
Thanks for the report.
Bad event mode behavior at default buffer size has been acknowledged (hence the buffer size is configurable - I wasn't confident that the 10ms was good for everyone).
I have no idea what the issue with 16bit files at high sample rates is - it does not make sense knowing how the plug-in works (it gets handed 32bit float data regardless of input bit depth, in fact it entirely does not care about input bit depth).