|
by Florian Grond, PhD
Content of the articleIn this blog entry, I will first give a brief overview of HOA, followed by an explanation of some relevant features of Dolby Atmos. Then, I will describe the general workflow connecting both approaches to create immersive audio experiences. I will conclude with some comments regarding the possibilities and limitations of some specific DAWs for the proposed workflow. Higher-order Ambisonics (HOA) overviewHigher-order Ambisonics (HOA) plays a key role in all situations that require isotropic sound scenes, i.e. scenes that uniformly cover each direction. These scenes comprise VR, AR, XR, video games and generally speaking 360° audiovisual content, for instance when streaming over YouTube or Facebook. The key advantage of HOA is the ability to transform and conveniently manipulate the sound scene in a uniform and mathematically rigorous way. Ambisonics started in the 70ies through Michael Gerzon’s work with what today is referred to as first-order Ambisonics (FOA) [1]. Higher-order Ambisonics was developed in the 90ies by various researchers [2]; the spatial resolution and the sense of depth in the sound scene improved significantly. Without explaining Ambisonics in detail here I will cover some aspects that are relevant to the workflow; if you want to know more, I have compiled more about HOA below [3]. The A- and the B-format It is important to know that the raw output of an Ambisonics microphone array, the A-format, needs to be converted into what is known as the B-format before using it in an Ambisonics production workflow. This is sometimes a bit of an obstacle for the novice. Unlike in a channel-based approach, the audio channels of the B-format are not associated in a direct way with spatial directions or positions, like the positions of speakers in the playback system, for instance. However, the more abstract representation of the sound scene in the B-format makes sound field operations like rotations fairly straightforward. Although multichannel audio is at its base, the B-format is only an intermediary step and requires decoding before you can listen to a meaningful output. This decoding process can either yield speaker feeds ranging from classic stereo to multichannel surround sound setups for a conventional channel-based output, or it can result in a binaural experience over headphones. This is just one of the reasons why one would like to deliver immersive audio recorded through Ambisonics over playback solutions that support Dolby Atmos. For delivering a mobile and immersive listener experience with comparably small hardware requirements, the capabilities of headphones with integrated head-tracking are particularly attractive. Ambisonic orders For FOA recordings, the A- and the B-format have both 4 channels. The spatial resolution is limited, and sound sources usually appear as if they are all at the same remote distance. For full-sphere HOA solutions the A-format has typically more channels than the B-format. As an example, the ZYLIA ZM-1 has 19 channels as raw A-format output and 16 channels after converting it to 3rd order Ambisonics. It is important to remember these channel counts when planning your workflow with respect to the capacities of the DAW in question. Dolby Atmos overviewDolby Atmos is a surround sound technology developed by Dolby Laboratories. On the level of channel counts and positions, it expands on existing surround sound systems by adding height channels. While Dolby Atmos as a format is agnostic with respect to the speaker layout of the playback system, the inner workings are in part inspired by a channel-based approach. Dolby Atmos is not isotropic as Ambisonics is, however, it does aim in a similar way to envelop the listener from many directions, hence the desire to explore links with the immersive quality of HOA recordings. In order to represent rich 360° audio content, Dolby Atmos can handle internally up to 128 channels of audio. The Audio Definition Model (ADM) ensures a proper representation of all the metadata related to these channels. Dolby Atmos files are distributed through the Broadcast Wave Format (BWF) [4]. From a mixing point of view, and also for combining it with Ambisonics, it is important to be aware of two main concepts in Dolby Atmos: beds and audio objects. Beds You can think of beds in two ways:
If you think for instance of a nature soundscape, this could be trees with rustling leaves and a creek with running water, all sound sources with more or less distinct positions. Take an urban soundscape as another example and think of traffic with various moving cars, these are sound sources that change their positions, but you would want to use the scene as is and not touch the sources individually in your mix. These are all examples of immersive audio content that you would send to beds. Audio objects Dolby Atmos also allows for sounds to be interpreted as objects with positions in three dimensions. In a mix, these are objects that allow for control of their position in x, y, and z independently of the position of designated speaker channels. See below [5], for reading up more on beds and objects. HOA recordings in Dolby AtmosThere are many ways how HOA sound material can be used in Dolby Atmos, involving both, beds and objects and the mixer is free to select or switch between those possibilities. For the scope of this article, let’s focus on mapping HOA to beds, in order to make a first step towards leveraging the immersive properties of Ambisonics with the end user format of Dolby Atmos. Mapping HOA to beds While beds are channel-based in their conception, they may be rendered differently, depending on the speaker count and layout of your system. Think of beds as your main mix bus and let’s think of input for beds as surround configurations (2.0, 3.0, 5.0, 5.1, 7.0, 7.1, 7.0.2, or 7.1.2). In order to take advantage of the high resolution of 3rd order recordings made with the ZYLIA ZM-1, we will pick the 7.0.2 configuration with 7 horizontal and two elevated frontal speakers and we will decode the Ambisonic B-format to a virtual 7.0.2 speaker configuration. This results in a proper input for a Dolby Atmos bed. The signal chain Starting with a raw recording made with the ZYLIA ZM-1 we will then have the following signal chain:  STEP 1 Step one is the raw output of the microphone array, the A-format. For the ZYLIA ZM-1 this is an audio file with 19 channels. From a post-production and mixing perspective, all that matters here is where you placed your microphone with respect to the sound sources. If you want your work environment to include this step of the signal chain, the tracks of your DAW need to be able to accommodate 19 channels. But this is not absolutely necessary, you can start with step 2. STEP 2 The 3rd order Ambisonic B-format contains 16 channels. For the conversion from step one to step two, you can use the Ambisonic Converter plugin from Zylia [6]. If your DAW cannot accommodate the necessary 19 channels for step one you can also convert offline with the ZYLIA Ambisonics Converter application, which also offers you batch conversion for multiple assets [7]. In many situations, it is advisable to start the signal chain with step 2, in order to save the CPU resources used by the A to B conversion for other effects. From a mixing perspective, operations that you apply here are mostly rotations, and global filtering, limiting or compression of the immersive sound scene that you want to send to Atmos beds. You will apply these operations based on how the immersive bed interacts with the objects that you may want to add to your scene later. There are various recognized free tools available to manipulate HOA B-format, for instance, the IEM [8] or SPARTA [9] plugin suites.  Zylia Ambisonics Converter plugin opened in Reaper STEP 3 Then, The Ambisonic B-Format needs to be decoded to a virtual surround speaker configuration. For this conversion from the B-format, you can use various decoders that are again available from multiple plugin suites like IEM and SPARTA. ZYLIA Studio Pro [10] allows you to decode to a virtual surround layout directly from step one, the raw A-format recordings, which means that you can bypass step 2. For some background audio content, this maybe a perfectly suitable choice. Part of the roadmap for ZYLIA Studio Pro is to also offer A-format input, making it a versatile high-quality decoder. From a mixing perspective and depending on the content of your bed input, you may want to choose different virtual surround configurations to decode to. Some content might be good on a smaller, more frontal bed e.g. 3.1, and other content will need to be more enveloping. If your DAW has a channel count per track that is limited to surround sound setups, you will need to premix these beds as stems.  Zylia Studio PRO plugin with 7.1.4 surround pre-set uploaded STEP 4 This bed then needs to be routed to Dolby Atmos. The details are beyond the scope of this article, and there are many excellent tutorials available that describe this process in detail. Here I want to mention that some DAWs have Dolby Atmos renderers built in, and you can study everything you practically need to know within these DAWs. With other DAWs, you will need to use the external Dolby Bridge [11]. This has a steeper learning curve to it but there are also many excellent tutorials out there that cover these topics [12]. There are also hardware solutions for Dolby Atmos renderings which interface with your speaker setup, but we will not cover them here. In Dolby Atmos, you will likely also integrate additional sources as objects, and you will control their 3D pan position with the Dolby Atmos Music Panner plug-in in your DAW. From a mixing perspective: the sonic interaction between the bed and the objects will probably make you revisit steps 2 and step 3 in order to rebalance, compress or limit your bed to optimise your mix. STEP 5 You will need to monitor your mix to make sure that the end user experience is perfect. Only very few of us will have access to a Dolby Atmos studio for their work. For bedroom studio owners, you can listen to your mix always over headphones as a binaural rendering, on some recent OSX platforms over the inbuilt Atmos speakers, and with AirPods even over headphones with built-in headtracking. These solutions might be options depending on what you are producing for. Regarding this highly debated question, on whether you can mix and master over headphones, I found the following article very insightful [13], elaborating on all pros and cons and also pointing out that the overwhelming majority of end users will listen to music over headphones. With regards to an Ambisonic mix, using headphones means that the listener will be always in the sweet spot of the spatial reproduction. The workflow in selected DAWsI will finish by discussing the steps of the signal chain on a couple of selected DAWs focusing on Ambisonics and beds only. In general, when planning your HOA to Dolby Atmos workflow, you need to understand how Dolby Atmos is supported in the DAW of your choice [14]. Either a Dolby Atmos renderer is integrated, or you need to get acquainted with the Dolby Bridge. On the side of HOA, you need to understand how many channels per track your DAW supports, and whether you can start working with the A-format, the B-format or premixed beds as stems. REAPER Reaper is amongst the first choices when it comes to higher-order Ambisonics, due to its 64-channel count per track. Hence for the HOA aspect of the workflow sketched above, there are no limitations. However, you will need to familiarize yourself with the Dolby Bridge and the Dolby Atmos Music Panner plug-in. PRO TOOLS In regular Pro Tools, you will also use the Dolby Atmos Music Panner plug-in and the Dolby Bridge. Since Pro Tools has a limitation of 16 channels per track, you will need to convert all your Ambisonic assets to B-format before you can start mixing. Upgrading to Pro Tools Studio or Flex [15] adds Dolby Atmos ADM BWF import/export, native immersive panning, I/O set-up integration with the Dolby Atmos Renderer, and a number of other Dolby Atmos workflow features as well as Ambisonics tracks. LOGIC In the most recent versions of Logic, Dolby Atmos is completely integrated, so no need to use the Dolby Bridge. For the monitoring of your mix, Logic will play nicely with all Atmos-ready features from Apple hardware. However, the channel count per track is limited to beds with 7.1.4. In theory, this means that you would have to premix all the beds as multichannel stems. While you can import ADM BWF files, as the Dolby Atmos project is ready for mixing, it is less obvious how to import a bed input as discussed above. In any case, once you have a premixed bed, the only modifications available to you in the mixing process are multi-mono plugins (e.g., filters), so you cannot rotate the Ambisonic sound field anymore at this point. To summarize for Logic, while Dolby Atmos is very well integrated, the HOA part of the signal chain is more difficult to realize. NUENDO Nuendo also has Dolby Atmos integrated and it also features dedicated Ambisonic tracks up to 3rd order which can be decoded to surround tracks. This means you have a complete environment for the steps of the workflow described above. DAVINCI RESOLVE While being mostly known as a video editing environment, DaVinci Resolve features a native Dolby Atmos renderer that can import and export master files. This allows for a self-contained Dolby Atmos workflow in Resolve without the need for the Dolby Atmos Production or Mastering Suite. DaVinci Resolve also has the Dolby Atmos renderer integrated and the tracks can host multichannel audio assets and effects. SummaryI hope this gives a general overview of how to connect HOA with Dolby Atmos. The proposed workflow is making use of beds in an orthodox way, but this is not the only way how HOA recordings can be mapped to Dolby Atmos. The 128 channels of Dolby Atmos including the object channels offer plenty of opportunities to explore. If you like this article, then please let us know in the comments what we should describe in more detail in future articles? References:[1] The publication by Michael Gerzon introducing FOA:
[6] The Zylia Ambisonics Converter plugin: LINK [7] The Zylia Ambisonics Converter: LINK [8] The IEM plugin suite: LINK [9] The SPARTA plugin suite: LINK [10] Zylia Studio Pro plugin: LINK [11] A video tutorial for using the Dolby Bridge with Pro Tools: LINK [12] A video tutorial for using the Dolby Bridge with Reaper: LINK [13] An blog post about the limits and possibilities of mixing Dolby Atmos via headphones by Edgar Rothermich: LINK [14] Information about Dolby support for various DAWs: LINK [15] Here you can compare Protools versions and their Dolby and HOA support: LINK #zylia #dolbyatmos #ambisonics
1 Comment
2/6/2023 08:26:57
It was a great experience while reading your blog post. I bookmarked and love to visit on a regular basis. Leave a Reply. |
Categories
All
Archives
August 2023
|
RSS Feed
|
Connect with us |
|
© Zylia Sp. z o.o., copyright 2018. ALL RIGHTS RESERVED.
|