About MeshRIR

MeshRIR is a dataset of acoustic room impulse responses (RIRs) at finely meshed grid points. Two subdatasets are currently available: one consists of IRs in a 3D cuboidal region from a single source, and the other consists of IRs in a 2D square region from an array of 32 sources. This dataset is suitable for evaluating sound field analysis and synthesis methods.

Download

The dataset is available in the following link:

• https://zenodo.org/doi/10.5281/zenodo.5002817

Extract the data files in the folder “src”. For processing the IR data, see “ir_view.html” (converted from .ipynb) for Python or “ir_view.m” for Matlab. Example codes for sound field analysis and synthesis are included in the folder “example”. Latest codes are available here:

• https://github.com/sh01k/MeshRIR

You can also try on Google Colab:

If you use the MeshRIR dataset for your research, please cite the following paper (preprint):

@inproceedings{MeshRIR,
  author    = "Shoichi Koyama and Tomoya Nishida and Keisuke Kimura and 
               Takumi Abe and Natsuki Ueno and Jesper Brunnstr\"{o}m",
  title     = "{MeshRIR}: A Dataset of Room Impulse Responses on Meshed Grid Points For Evaluating Sound Field Analysis and Synthesis Methods",
  booktitle = "Proc. {IEEE} Int. Workshop Appl. Signal Process. Audio Acoust. (WASPAA)",
  year      = "2021"
}

License

The MeshRIR dataset is provided under Creative Commons Attribution 4.0 International (CC BY 4.0).

Detailed description

MeshRIR dataset consists of two subdatasets.

S1-M3969

• IRs inside a 3D cuboidal region from a single source position

S32-M441

• IRs inside a 2D square region from 32 source positions

Source and microphone positions of S1-M3969 (left) and S32-M441 (right)

File format

The file formats are .npy for Python (Numpy) and .mat for MATLAB. All the additional data are provided as a JSON file.

Measurement conditions

The IR at each position was measured using a Cartesian robot with an omnidirectional microphone (Primo, EM272J). The signal input and output were controlled by a PC with a Dante interface. The loudspeaker was DIATONE, DS-7 for S1-M3989 and YAMAHA, VXS1MLB for S32-M441.

	S1-M3969	S32-M441
Sampling rate	48000 Hz
IR length	32768 samples
Room dimensions	7.0 m × 6.4 m × 2.7 m
Number of sources	1	32
Measurement region	1.0 m × 1.0 m × 0.4 m	1.0 m × 1.0 m
Intervals of mics	0.05 m
Number of mics	21 × 21 × 9 points	21 × 21 points
RT60	0.38 s	0.19 s
Avg. temperature	26.3 °C	17.1 °C

IR measurement system for S32-M441

Usage

Basic usage examples (Python)

See ir_view.ipynb for the details.

• Import irutilities.py

  import irutilities as irutil
  

• Load microphone and source positions, and IR data

  posMic, posSrc, ir = irutil.loadIR(path_to_data_folder)
  

• Plot IR of source srcIdx and mic micIdx

  irutil.irPlots(ir[srcIdx, micIdx, :], samplerate)
  

Basic usage examples (MATLAB)

See ir_view.m for the details.

• Add folder “matfiles” to search path

  addpath('matfiles');
  

• Load microphone and source positions, and IR data

  [posMic, posSrc, ir] = loadIR(path_to_data_folder);
  

Basic usage examples (Julia)

• Import irutils.jl

  include("lib/utils.jl")
  import .utils
  

• Load microphone and source positions, and IR data

  posAll, posSrc, irAll = irutils.loadIR(path_to_data_folder)
  

Application examples

• Sound field reconstruction: examples/sf_reconst.ipynb
  • Estimation of pressure distribution using the method proposed in [2].
  • Microphone positions are selected by the MSE-based sensor placement method [3].
• Sound field control: examples/sf_control.ipynb
  • Synthesis of planewave field by pressure matching.
  • Weighted pressure and mode matching [4,5] are also demonstrated. Details can be found in [6].

References

1. S. Koyama, T. Nishida, K. Kimura, T. Abe, N. Ueno, and J. Brunnström, “MeshRIR: A dataset of room impulse responses on meshed grid points for evaluating sound field analysis and synthesis methods,” in Proc. IEEE WASPAA, 2021. [pdf]
2. N. Ueno, S. Koyama, and H. Saruwatari, “Sound field recording using distributed microphones based on harmonic analysis of infinite order,” IEEE SPL, 2018. [pdf]
3. T. Nishida, N. Ueno, S. Koyama, and H. Saruwatari, “Region-restricted Sensor Placement Based on Gaussian Process for Sound Field Estimation,” IEEE Trans. SP, 2022. [pdf]
4. S. Koyama and K. Arikawa, “Weighted pressure matching based on kernel interpolation for sound field reproduction,” Proc. ICA, 2022. [pdf]
5. N. Ueno, S. Koyama, and H. Saruwatari, “Three-dimensional sound field reproduction based on weighted mode-matching method,” IEEE/ACM Trans. ASLP, 2019. [pdf]
6. S. Koyama, K. Kimura, and N. Ueno, “Weighted pressure and mode matching for sound field reproduction: Theoretical and experimental comparisons,” J. AES, 2022. [pdf]

Author

• Shoichi Koyama (National Institute of Informatics, Tokyo, Japan)
• Tomoya Nishida
• Keisuke Kimura
• Takumi Abe
• Natsuki Ueno
• Jesper Brunnström