VA Pilot Session - Mixing Implementation of VA into Logic Pro X with Unity Engine + Steam Audio
Running through the first implementation gained valuable insights on the recording process of the Steam Audio implementation in Unity engine. With the initial setup across Logic Pro X and Unity Engine sessions, the internal routing software Blackhole had led to a mismatch between the two project sessions due to unconfigured sample rate settings, leading to slowed audio in the engine and heavily distorted recordings in the DAW. Despite this the initial setup itself was quick and straightforward, with the CPU and memory demand being reasonable.
Setup on the engine session was quick, utilising some ready VAS geometry tagged scenes, source setup was quick with the multitracks imported. Here would have been a point where a later error and reset could’ve been avoided, as imported stem files came from the original set whereas if they had been imported into the Logic Pro session and then into the engine there would have been no further mismatch due to the inconsistent sample-rates.
However, once the setup of both sessions had been done using the internal routing software, signal was monitored from the engine session and confirmed to be coming into the DAW mixer on the metre. On monitoring the initial setup within the engine, to adjust the narrow sound some changes to the arrangement underwent until there was more distinction with height and depth in the front hemisphere. Image sources were pushed further to the diagonal axis, this resulted in a wider sound that captured reflections then generated from some of the acoustic geometry structure further within the VAS leading to a more complex spectra from the initial metallic sound. In the settings of Steam Audio I increased the ambisonics order, leading in more directional clarity and definition of depth.
As recording had not yet been tested, the setup continued with final arrangement and placement of image sources and HRTF receiver within the VAS. Within the engine itself, audio playback was clear but quality was noticeably lacking. Arrangements were set initially within the front of the perspective, but with a narrow stereo width resulting this was later adjusted. With the stereo signals requiring split mono signals to be placed in two engine image sources, these stems were exported from Logic Pro at this point in split mono and then reimported into the session and assigned to the LR space pair sources.
At this point, a test recording was done and the resulting noise informed me that there were significant errors. With the cyclic framework of research informing action, reflection that underscored this debugging stage was beneficial to resolve these occurring issues. When reflecting after the initial errors became apparent, it occurred that both the tempo change and distortion meant that the sample-rates had not been set consistently. Resetting at this point however with new DAW and engine sessions was surprisingly not a daunting prospect, due to the swift practicality of the initial setup which was an indicator for the efficiency of this setup when considering time saving on errors. Once the new sessions were set up and linked with corrected assigned sample rates across the DAW / Unity / Blackhole / Interface, I assigned the image sources with new stems and with a quick test of recording gave consistent audio to the real-time engine monitoring.
Further configuration was done following the reflection of several test iterations, then final recording of the demo was undertaken. These stems being already within the Logic Pro session once rendered and recorded, they were then mixed into the track for creative application and use.
The processing yielded good results, with the efficiency and implementation process being practical and effective. The interesting resulting quality of audio demonstrates potential for further development of higher definition audio source placement and virtual acoustic encoding, along with there being room to improve on the overall exocentric effect.
Following the pilot test session, practicality and efficiency is apparent in the implementation of engine rendering to DAW. Audio quality development of VA appears to have a high upper limit with the array of quality settings such as ray and reflection cue counts, as well as with creative use of source arrangement relative to listener perspective. Creative application itself lends to a high degree, with the combination of arrangement and acoustic geometry modelling being as varied as in real-world acoustic environments, if not more so with the ability of unconventional placement available. Reflection cue generation itself also lends to a versatility which mirrors real-world acoustic environment design, with the acoustic tagged geometry setting having noticeable impact on the final sound.
Mixdown of dry to VA stems, with automated crossfade between the two :
AutomatedFadeDryToVA-UnfinishedDreamsVADemoS2.wav
https://drive.google.com/file/d/1bWU3DNYcIMdRQNWpNbOL8cpqwzztzTwR/view?usp=share_link
