3DREAM

3DREAM: 3D-Printed Realtime Ambisonic Microphone

3DREAM is a collaboration between Emeraude and the Laboratoire de Mécanique des Fluides et d’Acoustique (LMFA) at École Centrale de Lyon. It is funded by the Fédération Informatique de Lyon (FIL). 3DREAM will be officially launched in January 2026. For any information, feel free to contact Romain MICHON (romain_dot_michon_at_inria_dot_fr).

Project Overview

Ambisonic microphones, used to capture sound in three dimensions, have become an essential tool in fields such as virtual reality, immersive cinema, and acoustic research [1]. However, commercially available solutions (e.g., Eigenmike 64 [2], Zylia ZM-1 [3]) are often expensive (~€1.5k to €15k), closed, and not very configurable, which limits their accessibility for both researchers and independent creators.

In this context, the 3DREAM project (3D Printed Real-Time Ambisonic Microphone) aims to design a DIY ambisonic microphone, 3D-printed, using low-cost MEMS microphones and an FPGA for real-time processing, thanks to the Syfala compiler [4], developed by the Emeraude team at CITI. 3DREAM will be carried out in collaboration with the Acoustics team at LMFA (École Centrale de Lyon), which will provide expertise in ambisonic methods.

The project will build on tools developed within these two laboratories: Faust [5] (Emeraude), Syfala (Emeraude), the SMALL library (Spherical Microphone Array Little Library) [6] (LMFA), and Ambitools [7] (LMFA). Faust is a domain-specific programming language (DSL) for real-time audio signal processing, developed at GRAME-CNCM in Lyon for over 20 years, and now in collaboration with the Emeraude team. It can be used with Syfala to program FPGA platforms at a high level for real-time audio signal processing applications. The SMALL library facilitates the design of 3D-printable ambisonic microphones. It offers a collection of modular CAD files as well as Faust code for processing signals from the capsules placed on the microphones. Ambitools is a collection of tools, also written in Faust, enabling the implementation of ambisonic encoders and decoders.

In this context, the two main objectives of 3DREAM are therefore to:

  • make ambisonics more accessible and adaptable;
  • improve Syfala by testing it in a concrete use case.

Methodology

The 3DREAM project will run over 2 years and will be structured around 4 tasks. One intern will be assigned to each task (2 funded by FIL and 2 funded through internal resources).

Extension of the SMALL library and implementation of encoding processes in Faust
This first stage will focus on expanding the SMALL library [6]. It will involve producing a variety of microphone array configurations (in terms of ambisonic orders, geometry, and number of capsules) to cover a wide range of use cases. Several encoding algorithms [8,9] for ambisonic signals will also be implemented in Faust. The goal is to build a rich software base, immediately usable for the rest of the project and easily integrable into the Faust → Syfala → FPGA compilation chain.

Improvement of the Syfala compiler’s Faust mode for multichannel processing
The second task will focus on enhancing the Syfala compiler to optimize its operation in Faust mode, in the context of low-latency multichannel audio processing. This will involve adapting the compiler’s architecture to better handle complex signal (input/output) configurations, automating parallelization and pipelining, and evaluating performance in terms of FPGA resources and latency. To this end, new optimization techniques [10] will be integrated into Faust. These developments will strengthen Syfala’s robustness and versatility in real-time spatialization scenarios.

Development of a 3D-printed ambisonic microphone prototype
This task will involve designing and assembling one or more prototypes of 3D-printed ambisonic microphones, integrating low-cost MEMS microphones and an FPGA board for real-time embedded processing. The aim is to provide a lightweight, mobile, and autonomous solution based on an open hardware architecture. Two modes of operation are envisioned:

  • a portable mode, with direct recording of processed signals onto an SD card,
  • a connected mode, enabling real-time audio data streaming via an Ethernet interface.

Particular attention will be paid to compactness, energy efficiency, and system robustness for use in varied conditions. Several technological challenges will need to be addressed in this task:

  • Managing the wiring and shielding of a large number of MEMS capsules (up to 32 in the most extreme case) concentrated on a small spherical surface;
  • Transmitting real-time audio streams via Ethernet without relying on proprietary protocols such as DANTE, etc.

For this last point, we plan to use an open protocol recently developed by the Emeraude team [11].

Acoustic characterization and experimentation in archaeoacoustics
The developed prototypes will undergo acoustic characterization: array calibration, performance in terms of signal-to-noise ratio, source localization, etc. Subsequently, these devices will be used in archaeoacoustic experiments, particularly within the AURACAVE project [11], which aims to explore the sound environments of prehistoric sites such as the Chauvet-Pont d’Arc cave. This application will allow the system’s performance to be tested in real conditions while contributing to the reconstruction of the sound environments of historical sites. From the captured 3D impulse responses, an auralization in the listening room of the École Centrale de Lyon will allow audiences to hear the acoustics of historical sites inaccessible to the general public—for example, by projecting their own voices into them.

References

[1] Zotter, F., & Frank, M. (2019). Ambisonics: A practical 3D audio theory for recording, studio production, sound reinforcement, and virtual reality (p. 210). Springer Nature.

[2] https://eigenmike.com/eigenmike-64

[3] https://www.zylia.co/shop

[4] Popoff, M., Michon, R., Risset, T., Orlarey, Y., & Letz, S. (2022). Towards an fpga-based compilation flow for ultra-low latency audio signal processing. In Proceedings of the SMC-22 Sound and Music Computing conference.

[5] Orlarey, Y., Fober, D., & Letz, S. (2009). Faust: an efficient functional approach to dsp programming. New computational paradigms for computer music.

[6] Lecomte, P. (2023, September). SMALL: Spherical Microphone Arrays Little Library. In Forum Acusticum 2023-10th Convention of the European Acoustics Association.

[7] Lecomte, P. (2018). Ambitools: Tools For Sound Field Synthesis Using Higher Order Ambisonics-V1. 0. In 1st international Faust Conference (IFC-18).

[8] Zotter, F. (2018). A linear-phase filter-bank approach to process rigid spherical microphone array recordings. In Proceedings of IcETRAN.

[9] Moreau, S., Daniel, J., & Bertet, S. (2006, May). 3d sound field recording with higher order ambisonics–objective measurements and validation of a 4th order spherical microphone. In Proceedings of the 120th Convention of the AES (pp. 20-23).

[10] Popoff, M. (2024). Compilation of real-time audio DSP on FPGA (Thèse, INSA de Lyon).[11] https://ccrma.stanford.edu/chauvet/