APL Virtuoso - the Ultimate Binaural Translator
by hklworld   14th April 2023

Hi John, on Pro Tools, you could first create a Dolby Atmos re-render file and then load it onto a 7.1.4 or 9.1.6 track, insert Virtuoso and then export the binaural stereo output as a wav.
Alternatively, you could insert Virtuoso in the live re-render track for the internal DAR, set a virtual audio driver like Blackhole 16ch as the PT output, and then route the final output into another DAW, like Reaper, to record and export live. The same applies if you use the external DAR - Virtuoso chain.

Here is how to use Virtuoso with the live re-render of Pro Tools internal DAR: https://youtu.be/S-5IDKFW_Sw?si=rVC7kb7JtPnIGLCt

I just bought Virtuoso a few days ago and it's amazing! I have been using the binaural monitoring feature of Davinci Resolve's internal Atmos renderer but this is on another level, especially when combined with webcam headtracking solutions! It's the first time I am convinced the sound seems to come from speakers in my room (even though I have none)! I like HRTF profile D but the center channel keeps giving me impression that it's at the rear, so I tried out the ones here:
https://sofacoustics.org/data/database/widespread/
The SOFA I ended up loading into virtuoso is "UV02m_00003", the positioning is less obvious than profile D, but it doesn't give my any weird phase issue at any spatial positioning, nor any wrong impression of positioning that doesn't match the panning in DAW.

(EDIT: After further digging into other databases, I found that "Subject44_HRIRs_dfeq" from https://sofacoustics.org/data/database/3d3a/ also works quite well with me)

I found the default volume is suitable for lower loudness content, but a lot of the time I found myself setting the gain to about -8 to -10 to avoid the output clipping, , and that gain setting keeps getting reset when changing room presets. Maybe there can be a feature to lock individual settings when changing presets?

Also, which headphone target curve is recommended? I have been using for some time now the -1 db/oct tilted bk5128 diffuse field (available on AutoEQ's web ui), I loaded my headphone's measurement from Crinacle (I double checked that they do use the BK5128 to measure the headphone, the same fixture as my target), and loaded it to AutoEQ's web ui, and use Equalizer APO to load the generated EQ filter.

Can I keep using the tilted diffuse field target? Or is it more preferred to use the OG diffuse field? Or does it have to do with the SOFA file I loaded? In that case, any info on what target curve to use with the "widespread" (EDIT: and the "3D3A") PHRTF database?

Does the auto channel mapping feature works in the standalone version? @ hklworld I am currently looking into the channel ordering confusion between programs (more on that here: https://sourceforge.net/p/mesh2hrtf-...er_in_Windows/), and I don't think the auto channel mapping feature is working in the standalone version, is it?

Any chance of having a "recalibrate" headphone profile to use Virtuoso 2 with SoundID Reference? Now that the Apollo add-on is out and sounds better than SoundID's own filters it would be interesting to hear what virtuoso sounds like with it.

+1 for retaining the settings when switching between room room types.

Been reading some past posts here. I am a bit perplexed on this. It seems everyone uses lower ambience, around 20%. I have been using Virtuoso heavily since I bought it a while back, mainly doing Dolby Atmos music. My current feeling is that, 50% is the baseline, going below that would make me unable to identify the distance between myself and the speakers, and if I want to make it sound super convincing, I will need to set it to somewhere between 70% and 100%, and the Living Room preset sounds the most convincing to me, it seems everyone else prefer low ambience while I keep needing more ambience to make it sound convincing.

I guess that has to do with me having never been to these heavily treated rooms, with all the sound-absorbing foam and what not. Can I assume that if I listen to my mixes in a treated studio, it will sound more or less like when setting the ambience to around 20%? Should I try to get use to lower ambience level around 20% or is it fine to just use higher ambience level between 70% to 100%?

We are working on the preset management system to make it possible to save custom room settings in a future update.

.
This is a very subjective thing, but the HPC EQ filters included in the list are mostly Oratory1990s, based on the Harman target. But different targets might work better depending on preference and how similar it is to actual speaker listening in the room. There are many variables here, and this is one of the reasons why we included the 10-band EQ. We are also considering extra bands for adding separate target curves.

Yes this depends on how the HRTF set you use are equalised. The default ones are diffuse field equalised, but different datasets use different ways of doing DFE. Also, when it comes to translation to speakers, people are familiar with different balances, and the headphone correction targets are also very subjective. So it's difficult to generalise anything really. But if you already found a specific SOFA file that works for you, that's a great start. I'd recommend that once you found the HRTF set and HPC EQ profile that work for you, then you could try fine-tuning the tonal balance using the 10-band EQ to match the tonal balance as close to reference speakers in your studio or a well-treated room as possible by comparing.

Sorry for the delayed reply. The auto channel mapping is for the plugin version only as it's related to DAW's specific channel mapping. Good point, we should remove it in the standalone.

Not sure if I understand what you mean by "recalibrate" profile. SoundID Apollo add-on is to integrate SoundID correction profile into UA's DSP. So you can listen to Virtuoso output with your SoundID calibrated headphone profile instead of Virtuoso's HPC EQ if you like.

Yes, we are planning to implement a preset management in a future version.

In this post you say Virtuoso and Sonarwork have different targets. What I'm asking is some sort of "translation" between the two by loading a HPC that compensates this difference. Like Realphones does.

APL Virtuoso - the Ultimate Binaural Translator

A quick update: in about two weeks time, we will be releasing v2.0.1 with a few bug fixes and some updated features:
- bug fixes on the ADM player (moving objects), HPC EQ file loading and headtracking drift correction with Maxwell.
- HPC EQ: added an enable/disable switch for each band.
- HPC EQ: added Export for saving custom settings in a txt file, so you can recall your own headphone EQ.
- Solo/Mute on the View tab: now simply a left click for Solo, a right click for Mute, and Shift-Click for X-OR solo/mute.

Ok, I see. But nobody seems to really know what Sonarwork's target curve is exactly. They don't publish their target unlike Harman does. It's supposed to be perceptually "flat" but this depends on what kind of ear simulator they use for measurement. That's very unclear either. To find out the difference, you can only see the difference by looking at how much correction it applies to a specific model. That's not the direction we'd go to. Virtuoso allows you to load a txt profile from AutoEQ which you can tweak easily, and there are many targets you can apply and try. And perhaps, an ideal solution is to speak with UA if they are willing to implement txt eq profile loading too. That should be very straightforward and most flexible.

This is an interesting and valid point. The reason why we added the ambience level control is that externalisation and the sense of realism really depends on how congruent the room you are physically in and the virtual room you are listening to in binaural reproduction are. Research suggests that if the congruency is poor, then the brain is confused and externalisation/localisation doesn't work instantly. Through training and continuous listening, this could be overcome to some extent, but requires a lot of cognitive load. There is also the effect of visual difference. The way our brain works is amazing, as soon as you walk into a room and hear a sound, then the brain learns the acoustic pattern and use it as a template, so for example, if the binaural sound is completely dry and you are in a very lively room then the sound won't externalise or appear very close to the head at max. But if you are in a dry space and listen to the binaural sound with a similar amount of reflections/reverb, then it will sound much more realistic. So it's natural that you'd prefer 70% or higher if your room is lively, and 10-20% if the room is a very heavily treated dry mix room. 50% for the APL listening room preset works perfectly in the real APL listening room as that's the amount of ambience in the room (people can't tell whether it's headphones or speakers), but if I use the same preset in my living room, then I need to customise it and turn it up to 100% to hear the sound exactly at the speaker position.

Can there be a feature to customize the speaker placement angle? I want to put the 9.1.6 surround speakers to 100 degrees while keeping the rear speakers at 135 degrees, but option A sets surround to 90 degrees and rear to 135, option B sets surround to 110 degrees and rear to 150. Would be more convenient to be able to customize the angle degrees.

Maybe this can one day be added to the future preset management system? The ability to store presets for channel mapping and speaker placement degree angle.

Plugin crashed when I was trying to open. It also causes my DAW to crash. How can I fix it?

We are aiming to add the room preset save/load feature in a future update. Making speaker positions more flexible is something we can look into for a future update as well.

Could you report it in the link here? with details of your DAW and OS, etc. Thanks.
https://docs.google.com/forms/d/e/1F...uhUZA/viewform

Alright I just submitted, also here is a link to the video: https://www.bilibili.com/video/BV1Pa...5327cfb8f3ab45

Thanks.

You could please add the new DCA Noire X to the list of headphones? I already own the Stealth but would love for anyone who has the Noire X (and the E3 if it hasn't already been added) to be able to benefit from the headphone calibration in this superb tech!

Did a bit of Googling, I found people sharing measurements done by Resolve (from Headphone.com), here: https://forum.headphones.com/t/dca-a...ussion/24663/3

I traced two of the measurement graphs (BK5128 and Gras), you can load into https://autoeq.app/ and generate the EQ.txt to load into Virtuoso (EDIT: Make sure to click "Show advanced" and choose your target! Do not EQ your headphone to flat! DON'T!) (just choose "EqualizerAPO ParametricEq" in the App section, the generated txt file works accross Apps, note Virtuoso seems to ignore the preamp, you might need to lower the gain manually to avoid clipping). Choose the graphs based on the target you want to EQ to, if your target is BK5128 ones use the 5128 measurement, if it's Harman 2018 over ear then use the Gras measurement.

Thanks for this!

I think I found a bug. I am using the standalone, the ADM player ignores the custom channel mapping, I usually swap the slots for 9.1.6 wide and rear channels, putting 9.1.6 wide surrounds on channel 9 and 10, but the ADM player still puts the rear channels to 9 and 10, I think it's ignoring my custom channel mapping.

Found another bug, after enabling the head tracking (using Open Track + nvsonic's Head Tracker OSC Bridge), what's supposed be on the top right rear area, sounds like it's on the rear area on the ground, I recorded a video for this one, I kept my head still facing front, the Yaw, Pitch, and Roll are all kept at 0 degrees, and I switch on and off the head tracking, ideally I shouldn't hear a difference, but the sound shifted places. It seems I am unable to properly hear the top channels when enabling head tracking.

Video here:
https://drive.google.com/file/d/1VGR...ew?usp=sharing
@ hklworld

I have been recommending virtuoso in some online chats, and some people actually tried it, they were happy with it but one guy complained about the mute/solo UI being not clear enough. He clicked on the speakers without awareness of what clicking would do, he then saw the speakers greyed out, he asked me how to fix it, I was still using an older version where right clicking brings up a "Solo/Mute" pop up menu, so I told him to right click and undo the soloing or muting. He then right clicked on different speakers, but the speakers are still not making sound (of course now we know that he muted the speakers by right clicking them).

He took a screenshot (https://message.biliimg.com/bfs/im_n...a325595004.png), in his screenshot, the three front speakers of L, R, and C are displayed normally, other speakers are greyed out. But he couldn't hear anything when playing stereo audio. I told him to check the Matrix page, he then saw that only the C speaker is on, everything elso has been muted.

This showed me that Virtuoso's new left and right clicking for Solo and Mute, still lacks visual cue for telling the user "what exactly happened when I clicked", the greying out cannot differentiate whether it's muted or soloed or both, the greying out sometimes is also bugged (he only had the center channel turned on, and yet all three front speakers are highlighted).

I would suggest add some visual cues to the speakers to indicate whether they are soloed, or they are muted. Maybe paint the speakers green when it's soloed, and paint them red when speakers are muted. Then when the mouse hover over the speakers, show a tooltip that says "This speaker is soloed/muted, left/right click on it to undo" or something. Not sure what to do if it's both soloed and muted, maybe paint it yellow? Or maybe just prevent that from happening, if it's soloed and the user right clicked on it, rather than having the speakers be both soloed and muted (which doesn't really make sense anyways), have the speakers be only muted instead; vice versa, if the speaker was muted in the first place, and then the user left click on it, instead of both solo and mute at the same time, have the speaker be only soloed and unmuted. If the user really want to both solo and mute a speaker they can do it in the Matrix page. That makes the most sense to me.

Thanks for recommending Virtuoso and for the suggestion. You read our mind : ) We have been working on adding visual cues for solo/mute as well as some other requested features for v2.2. We are aiming to release in a few weeks.

An interesting info, from Virtuoso's manual, we know that HRTF set D is B&K HATS from Meta's dataset, but I was wondering which HATS exactly it is (4128 or 5128), so I emailed Meta's contact, I got the answer that it's 5128 (which I believe is so far the only dummy head mic that has a realistically modeled human ear canal, instead of a 711 standard straight tube). Maybe not important, but interesting to know.

I uploaded two python scripts for generating diffuse field frequency response from SOFA file, and applying parametric EQ filters to SOFA files, I have successfully "diffuse field equalized" some of the un-equalized SOFA files and load them into Virtuoso. Maybe this will be helpful. https://github.com/EaryChow/HRIR-SOFA-Equalizer

Hi all,
We have updated VIRTUOSO to v2.2.0! Existing users can download it from My Account (https://apl-hud.com/account). VIRTUOSO is currently available at a 50% discount https://apl-hud.com/product/virtuoso/). The new features and bug fixes are as follows:

New
– Room settings can be copied to a Custom user preset slot which is then saved with the session: Ambience, RT60, Distance, Gain, Room dimensions, Room damping and Reverb liveliness.
– Above parameters can now be exported / imported as a JSON file.
– Display the Solo / Mute state of loudspeakers using floating icons in the View page.
– Added ability to clear loudspeaker solo and mute states from the View page.

Fixed
– Fixed crash when attempting to open Virtuoso on an M4 based Mac (all versions).
– Improved the loading times of ADM files in the Standalone version.
– Improved the audio processing stability of the Standalone version.

great stuff! when is virtuoso going to get that custom head circumference feature?