Artikel von Felix Andriessens
Ambience recordings are a crucial tool for designing acoustical environments and scenographies and thereby transporting emotions, for example in applications as film sound, game sound, VR applications, theatre and radio plays, to name a few.
Most ambience libraries currently available contain stereo recordings or surround ambiences built from stereo recordings; only some are actually recorded using surround microphone techniques. With the advent of playback solutions as Dolby Atmos, standardized speaker setups not only containing more than just two surround channels but also containing ceiling speaker channels have become of interest. The continuing boom of VR/AR/MR/360°-media of all kinds is leading to a raised demand for high quality 3D-audio stock material like eg. ambiences. Especially since sound never is the first concern during principal photography, no matter how much attention a dedicated sound recordist pays to it – there is always the need for material additionally to the location sound to achieve its optimal impact, even when producing documentaries. And while building 3D sound from mono and stereo sources is a valid way to repair, extend and/or replace an existent production sound, it becomes a laborious task when it comes to creating several layers of ambiences per scene as it needs to be done for feature films for example.
Our plan is to fill the gap between demand and supply especially for cinematic ambiences, and so our first aim was to develop a 3D-audio microphone setup that offers a large and stable sweet spot even when played back in large venues as cinemas, as well as proper channel separation especially between the lower and the upper speaker layer, to avoid having sound sources coming from the ceiling that really belong into the lower speaker layer.
In Summer 2016, we compared six 3D-audio microphones or microphone setups in an ambience recording session:
- three first order Ambisonic microphones
- a Schoeps ORTF-3D
- a combination of "Williams Star” (or “Williams Tree”, as we call it) and IRT cross (making it a 5.0.4 array)
- and our own development, a 7.0.2 microphone array called “Andriessens-Hoffmeister-Array”
- two prototype microphones not further mentioned here
The recordings have been mastered (meaning equalizing the levels, but not equalizing the sound of the different microphones in the test field) at Rotor Film in Potsdam Babelsberg, Germany - one of the largest Dolby Atmos mixing stages in Europe.
Since then, three listening tests have been conducted, two of them rather informal (at Rotor Film and at The Post Republic, Berlin, Germany, both with less than ten participants, all working as film sound designers and/or film re-recording mixers) and one being a blind (but not double-blind) test, the latter at Hamburgs University of Applied Sciences' (HAW) Tonlabor during Klingt Gut! Symposium on Sound 2017, involving more than 30 participants. All three tests resulted in an audience preference ranking with the AH-Array perceived as the best system and the two other stereophonic systems with time of arrival differences following (ORTF-3D, Williams/IRT). The First Order Ambisonics microphones were perceived as worse. For more detailed information please see our paper „A New 3D Microphone Array for Cinema, VR and Games – a Comparison“, International Conference on Spatial Audio (ICSA) 2017, Graz, Austria.
The pictures show:
1. (above) the complete comparison setup, with the AH-Array closest to the camera and its upper microphone layer out of the frame, followed by a Sennheiser Ambeo directly next to the AH-Arrays center microphone, both under a Schoeps ORTF-3D, followed by a combination of a 5-channel "Williams-Star” and an IRT-Cross and closing with three more Ambisonic microphones (Core Sound TetraMic, Soundfield ST350); all spread over an area of approx. 7m length. The height of the AH-Arrays' upper microphone layer (which brings it out of the frame here) is usually smaller than in this picture.
2. a shot of the recording location facing North, taken two days after the session (the camping van did not park at that spot during the session)
3. a shot of the recording location facing South, also taken two days after the session
4. a detail shot of the comparison setup, showing the AH-Arrays right microphone, the Williams Star / IRT Cross, the Soundfield ST350, and last but not least the Core Sound TetraMic (from left to right).
5. a detail shot of the comparison setup, showing the AH-Arrays L/C/R-microphones, the Sennheiser Ambeo next to the AH-Arrays center microphone, and the Schoeps ORTF-3D above them.
6. and 7. give an impression of the recordings setup we used: six Sounddevices 788 recording 48 channels, all linked via C.Link, and a laptop recording another 30 channels.
8. a shot from the mastering session at Studio F at Rotor Film, Potsdam, Germany, to give an impression of the rooms size.
A Google Earth shot to give an impression of what the area looks like - please note the Günter-König-sports ground in the Northern West of the recording location, as well as the Urbanstraße in the North and the Hasenheide/Gneisenaustraße in the South of it, the latter including a subway line, all of which you can hear in the recordings.
AUDIO Download: this is an edited summary (in terms of summarizing some of the most significant moments of the recordings) of the following microphones or setups:
• Sennheiser Ambeo B-format
• Sennheiser Ambeo rendered
• Core Sound TetraMic B-format
• Core Sound TetraMic rendered
• Schoeps ORTF-3D
• one “fake” 3D-ambience track built from the ORTF-3D's lower front L/R-channels as an example of how most of the 3D ambiences are currently built by sound designers and sound editors.
All recordings (besides those in B-format) have the following channel assignment, following mostly Dolbys convention for Dolby Atmos:
L = left
C = center
R = right
LFE = LFE (empty)
Lss = left side surround
Rss = right side surround
Lsr = left surround rear
Rsr = right surround rear
Tls = top left surround (= ceiling front left; the equivalent of Dolby Atmos' Lts-channel)
Trs = top right surround (= ceiling front right; the equivalent of Dolby Atmos' Rts-channel)
Tbl = top back left (our “own” designation since not included in Dolby Atmos = ceiling back left)
Tbr = top back right (our “own” designation since not included in Dolby Atmos = ceiling back right)
- All Ambisonic microphones have been rendered from A-format to B-format using their custom filter apps and curves coming with them; also, there always is a second version rendered from B-format to 7.0.4 using the free Soundfield plug-in. There are tools that might sound better and offer more options, but (at the time of testing) the Soundfield plug-in was or still is free.
- Since ORTF-3D and Williams-Star-IRT-Cross have less than seven channels on their “base-layer”, we copied the surround channels to rear and side surround while attenuating them to fill the surround without raising its overall level.
- Since the Andriessens-Hoffmeister-Array has been developed with Dolby Atmos in mind, it has only two ceiling channels (although it could easily be extended to recording four channels if needed). To make it sound the way it is supposed to be, the recordings we’re sending here should be played back with only two ceiling speakers top down / overhead, or even better using a Dolby Atmos speaker system and assigning the ceiling channels as objects, not including them in a “bed”. As we copied the front ceiling channels to the rear ceiling channels and attenuated them by 6dB for easier comparison here, you would then have to raise the levels of those two channels again accordingly.
- All files are mono BWAV including TC stamps for easier alignment.
Thanks to Christian Lerch at Dolby, Ulli Scuda, Andreas Turnwald, Dimitrios Kosmidis at Frauenhofer Institut, Łukasz Januszkiewicz, Anna Czerwoniec and Tomasz Żernicki at Zylia, Jürgen Breitlow and Martin Schneider at Sennheiser/Neumann, Martin Frühmorgen, Holger Lehmann, Christoph Engelke, Konstantin Jaenecke, Thomas Neumann, Gregor Bonse and Andi Drost at Rotor Film, Svein Berge, André Zacher, Jörg Theil, Jörn Nettingsmeier and last but by far not least Helmut Wittek at Schoeps for their help and support! As well as to Mark Mattingly for accompanying us on several occasions and taking great pictures (here: fig. 1, 9, 10 and 11).