A list of publicly available acoustic/room impulse response (AIR/RIR) datasets. Currently this repo only tracks real-world recorded room impulse responses. Datasets included here may contain other data (e.g., speech, noise) in addition to RIRs. But this repo only describes the datasets about their RIR contents below. While there may be other webpages (if not outdated) that list some of the RIR datasets, this repo provides a brief description about the content of each dataset so you don't have to download and inspect them one by one before realizing it is not what you desire.
This repo also provides some scripts for downloading datasets if possible. Some dataset may require registration to download, which we may not provide download links directly. Only rows marked with a ✔️ are directly downloadable by running ./get_all_rirs.sh [destination folder].
For getting spatial RIRs in ambisonic format (i.e., .sofa), please refer to another dedicated website.
| Name | Content | Year | Paper/Document | In script |
|---|---|---|---|---|
| GTU-RIR | GTU-RIR dataset, comprises more than 15,000 room impulse responses (RIRs). To collect this dataset, we developed a semi-automated sound recording system, and the entire process of constructing this system and extracting the RIRs is thoroughly explained in our GitHub repository(https://github.com/mehmetpekmezci/gtu-rir) | 2024 | GTU-RIR - Gebze Technical Universiry Room Impulse Response Dataset for Acoustic Learning | ✔️ |
| MIRACLE | The MIRACLE dataset contains a total of 856,128 single-channel impulse responses acquired across four different measurement scenarios with a planar 64-channel microphone array. A regular grid of 64 × 64 source locations was sampled for two different source plane to microphone array distances. The dataset also contains measurements on a densely sampled 33 × 33 grid for the short distance, as well as measurements with the presence of a reflective panel. | 2023 | MIRACLE - Microphone Array Impulse Response Dataset for Acoustic Learning | ✔️ |
| SoundCam | It includes 5,000 10-channel real-world measurements of room impulse responses and 2,000 10-channel recordings of music in three different rooms, including a controlled acoustic lab, an in-the-wild living room, and a conference room, with different humans in positions throughout each room. | 2023 | SoundCam: A Dataset for Tasks in Tracking and Identifying Humans from Real Room Acoustics | ✔️ |
| Arni | 132,037 RIRs measured using 5342 configurations of 55 acoustic panels in the variable acoustics laboratory Arni at Acoustics Lab of Aalto University. | 2022 | Calibrating the Sabine and Eyring formulas | |
| Motus | 3,320 higher-order Ambisonic RIRs measured using an Eigenmike em32, 4 loudspeaker positions, and 830 furniture configurations inside a single room at Acoustics Lab of Aalto University. Spherical photographs and 3D models are available for each measurement. | 2021 | A dataset of higher-order Ambisonic room impulse responses and 3D models measured in a room with varying furniture | |
| dEchorate | 1800 annotated RIRs obtained from 6 arrays of 5 microphones each, 6 sound sources and 11 different acoustic conditions. | 2021 | dEchorate: a Calibrated Room Impulse Response Dataset for Echo-aware Signal Processing | |
| MeshRIR | 4410 mono RIRs recorded in very dense grids in a moderately reverberant room with accurate coordinates. | 2021 | MeshRIR: A Dataset of Room Impulse Responses on Meshed Grid Points For Evaluating Sound Field Analysis and Synthesis Methods | |
| BUT Reverb Database | 1300+ mono channel RIRs recorded in 8 rooms | 2019 | Building and evaluation of a real room impulse response dataset | ✔️ |
| MIT IR Survey | 271 mono channel RIRs all recorded in distinct places | 2016 | Statistics of natural reverberation enable perceptual separation of sound and space | ✔️ |
| ACE Challenge | {1,2,3,5,8,32} channel RIRs recorded in 7 rooms | 2015 | The ACE challenge — Corpus description and performance evaluation | |
| Multichannel Impulse Response Database | 234 8-channel RIRs recorded in the same room with 3 levels of reverberation and different microphone array spacings | 2014 | Multichannel audio database in various acoustic environments | ✔️ |
| REVERB Challenge | 24 8-channel RIRs recorded in small, medium, and large rooms | 2013 | The reverb challenge: A common evaluation framework for dereverberation and recognition of reverberant speech | ✔️ |
| OpenAIR | Ambisonic B format RIRs recorded in over 46 (still increasing) environments | 2010 | Openair: An interactive auralization web resource and database | ✔️ |
| C4DM RIR database | 468 mono or ambisonic B format RIRs recorded in 3 large environments | 2010 | Database of Omnidirectional and B-Format Impulse Responses | |
| Aachen impulse response database | 344 binaural RIRs measured with a dummy head in 5 environments including a church | 2009 | A Binaural Room Impulse Response Database for the Evaluation of Dereverberation Algorithms | ✔️ |
| RWCP Sound Scene Database | 143 multi-channel RIRs recorded in 14 rooms | 2000 | Sound scene data collection in real acoustical environments | ✔️ |
Collecting real-world impulse responses is not easy. This repo tries to keep up with new real-world datasets available through volunteers. If you know publicly available datasets that are not listed here, you are welcome to create a pull request by adding a new entry to the list (and optionally the script to download & organize it). When writing the description, please highlight how many IRs and environments and what channel format are invovled in this dataset.
Please properly cite whichever datasets you use in your paper. It would be great if you can also provide a link to this page (e.g., as a footnote, not a citation) so that others may find it useful.
