VA Mixing Session - What Will Happen ?

In this session some novel placements of audio sources will be utilised to split single stems of instrumentation up into varying positions around the stereo field. For this I will be first using a piano stem from an in-progress production remix track - What Will Happen? - there is the aim to discover some distinct placements working off of new experimentation informed by previous implementation research sessions.

Starting with the piano, the original stem [with EQ / Compression] was split into four layers each corresponding to a register across octaves for bass / midrange / upper midrange / top end presence. These four now split can be moved around the VAS independently of one another, allowing for added movement within the stereo field. With these 4 splits, to mimic the arrangement of a pianos internal strings from left bass to right highs, the following render was conducted to determine the timbral quality of this arrangement style.

Iteration 1 Render Recording : Keys Render 1 WWH.wav https://drive.google.com/file/d/1ngMsygKGCQ8g7viSW5VdC3dqHuy3l8wB/view?usp=share_link

The directionality here lends to a distinct placement which differs from conventional stereo placement as the audio can be perceived to move across the width of space in front and to the sides of the listener's perspective. The result is a diffuse sound which then is filled with reflection cues from the mid-sized VAS studio model.

For the next placement test, I kept the left to right placement of each register and added height for each step up in frequency band. This was then mirrored and flipped to create an X shape which crosses over so that both sides of the arrangement contain all 4 registers of the stem.

Iteration 2 Render Recording : Keys Render 2 WWH.wav https://drive.google.com/file/d/10qkv03436X9G1IajowA7V26VT1_majkn/view?usp=share_link

This render resulted in a mix of the two side placements summed centrally in the stereo field with somewhat minimal width and movement. The placement of mirrored sources such as here in this way may lead to a breakdown of ILD / ITD as the sources encounter one another at the same time on render at the HRTF, effectively losing the encoding properties to phase relationships which cancel one another out. The slight metallic sound has returned as an indicator of this merge of LR placed signals, whilst the overall stereo image doesn’t stray from the centre generally. Some reflections can be heard from side positions, which may be due to the non symmetrical placement of sources within the VAS itself leading to cues with varying ILD / ITD on either side, however the final sound does not achieve the aim of criss-crossing registers based off of source placement. Before moving to another placement, to see if these phase relationships can be amended whilst still keeping the LR signals all on each side, a minor adjustment in height to the right to left set of 4 has been made to shift them up in height so that they alternate relative to the left to right.

Iteration 3 Render Recording : Keys Render 3 WWH.wav https://drive.google.com/file/d/1LXNhKF4r0FyLmlfKb0xByUu6O_wzvIJo/view?usp=share_link

Upon completion of this iteration, it is noticeable that this alteration in height neither did not have the intended effect of splitting the two LR sides distinctly to contain stems occurring simultaneously on the other side. The stereo image remains mostly central, and there is unwanted and apparent crackling occurring due to the source count.

Returning then to the placement of just 4 sources, whilst keeping the addition of incremental height increases, the HRTF receiver (camera icon) has had its perspective rotated and height also adjusted to an optimal mic position for a balanced and cohesive final render. This form of audio modelling imparts timbral effects more prevalent than any other method in the sense that tonal shifts occur more frequently when exploring around the space for HRTFs receiver mic positions. This is similar to mic placement practice in real-world acoustics, as the placement of microphone has a significant impact on recorded tonality and timbre of any given instrumentation along with it’s recording space. Exploring mic placements rather than source placements may lead to some more concrete results of dynamic sound with omnidirectional properties whilst still maintaining full and lush harmonic and timbral qualities.

Therefore with this in mind, the following skewed receiver perspective with height adjustments leads to a much more balanced and dynamic sound which, which still maintains the proper exocentricity and directionality. Whereas previously the HRTF perspective was directly forward from the screenshot perspective, the white lines now seen indicate the rotation to the right in placement.

Iteration 4 Render Recording : Keys Render 4 WWH.wav https://drive.google.com/file/d/1uF9KC01NNuyclYaSBff6ef4oi2z3IsIF/view?usp=share_link

Continuing on, I followed on from the intro piano renders with the additional accompaniment layers of brass and strings. To begin with the brass, the stereo split stems were exported and positioned in a large reflective cathedral model to create a widely diffused VAS with reflections that would have significant ITD to draw out transient tails.

The LR signal sources were positioned in-line with the HRTFs receiver at far distances either side, with these signal sources identical in distance to the receiver. Each LR mono side of the original stereo signal was assigned to the direction of each LR input of the receiver, therefore no phasing occurs as the original stereo image is preserved albeit spread much wider. With wide positioning in the hard left / hard right perspective of the receiver as cues came mostly from these wide angles, nearer-field / vertical information was lacking therefore an ambisonic stereo source was positioned directly above the listener perspective.

This then meant the total signal playback count was four channels from three sources, to one HRTF listener, 8000 ray cues (assigned setting) with a total of 4 refractions each to impart additional incoming signal for acoustic reflection modelling and rendering. The distances of these three sources were adjusted, along with the listener position to find the best placement for a full timbre which emphasised the stereo depth. The model for the VAS was picked as the interior would make use of the refraction count, with the angled surfaces leading to rays casting in a diffusive environment

Audio render recording : Brass VA STEM What will happen? M2VA.wav
https://drive.google.com/file/d/11n3TBs_ag8idCgD1dVqgXcf2qIlamTZB/view?usp=share_link

For the next layer of instrumentation, the strings arrangement consisted of pitch-shifted layers from a tape sample. There being 5 individual stereo stems in total for this arrangement, these being - Low swells / original sample pitch / mid pitched / high pitched / all of the above together. For these I wanted to fill the stereo image similarly to the brass with wide positioning relative to the listener perspective.

For this however, along with the many stems, there was a need to ensure phase relationships were maintained so that the width was present in the final recording whilst both LR were balanced together. The model used for this render was chosen as it is a mid sized VAS which has a detailed concave surface on one side where positioning will go, whilst having an open / detailed space on all other angles. This should result in dense reflections from the detailed surfaces around sources whilst keeping an open air / low level to the opposite side to emphasise the sideways directionality.

After firstly trying a number of arrangements which resulted in phasing or lack of directionality, the most appropriate placement was to zig-zag the positioning of sources with the high register stems corresponding to the high positioned sources from the listener perspective and so on with the mid range positioned below, following with the original and all together layers sitting then above the low swell layers that were positioned at the base. With this arrangement setup on one side of the listener perspective, the reason then for the zig-zag on the axis parallel to the listener (Red arrow in figure) was so that as each left and right side were recorded independently, with a 180° rotation of the HRTF listener perspective the source positioning on either side the renders would be flipped resulting in no phasising from symmetrical stereo source placement. Each source utilised the ambisonic source component meaning the stereo signals would maintain it’s original LR image & phase relationships.

The result is two stereo signal renders for both LR sides, these recordings then when layered together in the DAW resulted in a maintained wide and deep stereo image which is panoramic, and imparting additional virtually acoustic raycast cues enveloping around the listener's perspective.

Audio render recording : LR Strings - What will happen? M2VA.wav
https://drive.google.com/file/d/15P4Et45wumOJSIWUeUM2OrqZR1b5Popo/view?usp=share_link
Left Side - L Strings VA STEM What will happen? M2VA.wav
https://drive.google.com/file/d/1tIyTbr1ZAZCu97MlmeKxjOsfyq3_f_1n/view?usp=share_link
Right Side - R Strings VA STEM What will happen? M2VA.wav
https://drive.google.com/file/d/1kZqXw2TXXwOsdrf-0I8nXcG9KiTYdGTe/view?usp=sharing
Audio render recording : Guitar VA STEM What will happen? M2VA.wav
https://drive.google.com/file/d/1E2jAgwFWX1PeQxsLH8FqhTULuZOE7s_6/view?usp=share_link
Audio render recording : Vox Outro VA STEM What will happen? M2VA.wav
https://drive.google.com/file/d/1_YLrV3TEbMOqxFgg7WuTQJgkKv5dQkge/view?usp=share_link
Vox Intro VA STEM What will happen? M2VA.wav
https://drive.google.com/file/d/1uxdKDH4GBjrO7NBS7010F4vDBAPqt-kI/view?usp=share_link