Why is the DMS3DS designed as an equally segmented multichannel 3D-audio microphone, array? Traditional multichannel microphone arrays are designed to cater the specific need of an application or listening environment. The most widespread surround speaker system have been developed in a time when stereo was the dominant standard and therefore a backward compatibility needed to be ensured. Although these standard have been widely adopted by many professional facilities, they could not find their way into most consumer homes. These systems are relatively costly and difficult to set up properly. In 2023, we are in a situation where most consumers experience immersive audio with their headphones thanks to technologies like ambisonic decoders and binauralization. This highly adoptable, versatile and relatively cost-effective solution has proven to be the most successful throughout many consumer applications. My approach is centered around the idea that the development of a great sounding 3D-audio microphone array should not be dictated by old playback standards which have not been widely accepted by consumers. Instead, I focussed on capturing the 360° sound field in the most natural sounding way possible, and therefore I choose a channel based approach using wide distanced omnis arranged in equal segments to cover the hole 360° surround sound field equally. This also allows some flexibility in post, since the image can be freely rotated without getting any phase issues. My philosophy behind the design of these arrays is to capture the best set of audio data possible and use new technologies to be able to create the best sounding experiences for new and well established media formats. This meant to go the extra mile and build this large array to be able to capture all nuances of an acoustic environment, that an ambisonic microphone will never be able to. The problem with ambisonic microphones is, that they are basically point source arrays. This means that the dataset will always be poorer than using widespread omnis. Virtual channel separation is mainly based on level differences. This leads to a relative small listening angle and lack of acoustic information and sense of space. But there is still a problem with equal segment surround arrays when you try to play the recordings back on traditional surround speaker setups. In the next section, I will describe the problem and offer a solution.
This is a brief introduction on how to prepare a 5.0 surround file (the format I ship the files in the DMS3DS ambience libraries) recorded with an equal segment microphone array to playback with a surround speaker setup recommended in ITU-R BS.775-3, which has become an industry standard adopted by many studios and audio-post facilities.
ITU-R BS.775-3 is basically a compromise to recommend a 5.1 surround speaker layout with locations that would work in a range of different scenarios across the many applications in broadcast, cinema and music recording/mixing. Such a 5.1-channel system would be adequate for most applications, but not the ideal solution for any of them. The front and left right speakers are placed left and right from the screen with a recommended loudspeaker base with of 2-4 m, positioned at ±-30° from the listening position to ensure compatibility with the well known 2-channel stereo format. In addition, a center speaker is added in the center at 0° from the listening positioned slightly behind the front L/R speakers so that they have the same distance to the listening position. The surround speakers are at the same listening distance, positioned between ±100-120° slightly behind the listener. This represents a compromise between the need for effects panning behind the listener and the lateral energy which is important for good envelopment. They are positioned more like side loudspeakers than like rear speakers, which can make it difficult to place sounds in the rear of the 360° sound field. To remedy this issues, the guideline recommends the extension to a 7.1-channel system and place the side speakers at ±60° and the rear speakers at ±150°.
There are two main problems here:
1) Local compressions and squeezes in the 360° surround sound field, leading to temporal compression and extension.
2) Poor rear (poor sense of envelopment) and lateral localization (hole in the side).