The 1st Clarity Enhancement Challenge

Clarity challenge code for the first enhancement challenge (CEC1).

Please visit the Clarity Challenge website for CEC1 information, and the Clarity Workshop website for CEC1 results.

Data structure

To download data, please visit here. The data is split into two packages: clarity_CEC1_data.train.tgz [192 GB], clarity_CEC1_data.dev_eval_metadata.tgz [163 GB]. Please also download and unpack clarity_CEC1_data.anechoic.v1_3.tgz [11.4 GB], which contains the correct version of anechoic signals for reference, and replace the old anechoic signals within the train.tgz and dev_eval_metadata.tgz.

Unpack packages under the same root directory using

tar -xvzf <PACKAGE_NAME>

Train contains the training data:

clarity_data
|
└───train
    └───interferers
    |   |    nosie 3.9G
    |   |    speech 4.5G
    |
    └───rooms
    |   |    ac 48M
    |   |    brir 46G
    |   |    rpf 379M
    |
    └───scenes 166G
    |
    └───targets 2.8G

Dev_Eval_Metadata contains development set, evaluation set (eval2 is the processed evaluation data by the baseline), and metadata,

clarity_data
|
└───dev
|   └───interferers
|   |   |    nosie 587M
|   |   |    speech 1.4G
|   |
|   └───rooms
|   |   |    ac 20M
|   |   |    brir 20G
|   |   |    rpf 158M
|   |
|   └───scenes 72G
|   |
|   └───targets 1.3G
|
└───eval
|   |   |    nosie 675M
|   |   |    speech 1.3G
|   |
|   └───rooms
|   |   |    ac 12M
|   |   |    brir 12G
|   |   |    rpf 95M
|   |
|   └───scenes 58G
|   |
|   └───targets 749M
|
└───eval2/scenes 21G

Data preparation

In this folder, we provide the code for generating train & scenes. If you simply tends to use the CEC1 data, please download with the link above.

Baseline

In the baseline, the baseline enhancement code using OpenMHA is provided. The evaluation code using the Cambridge Auditory Group MSBG hearing loss model and MBSTOI is also provided.

Refernces

• [1] Kayser, Hendrik, et al. “Open community platform for hearing aid algorithm research: open Master Hearing Aid (openMHA).” SoftwareX 17 (2022): 100953.

• [2] Baer, Thomas, and Brian CJ Moore. “Effects of spectral smearing on the intelligibility of sentences in noise.” The Journal of the Acoustical Society of America 94.3 (1993): 1229-1241.

• [3] Baer, Thomas, and Brian CJ Moore. “Effects of spectral smearing on the intelligibility of sentences in the presence of interfering speech.” The Journal of the Acoustical Society of America 95.4 (1994): 2277-2280.

• [4] Moore, Brian CJ, and Brian R. Glasberg. “Simulation of the effects of loudness recruitment and threshold elevation on the intelligibility of speech in quiet and in a background of speech.” The Journal of the Acoustical Society of America 94.4 (1993): 2050-2062.

• [5] Stone, Michael A., and Brian CJ Moore. “Tolerable hearing aid delays. I. Estimation of limits imposed by the auditory path alone using simulated hearing losses.” Ear and Hearing 20.3 (1999): 182-192.

• [6] Andersen, Asger Heidemann, et al. “Refinement and validation of the binaural short time objective intelligibility measure for spatially diverse conditions.” Speech Communication 102 (2018): 1-13.

Citing CEC1

@inproceedings{graetzer2021clarity,
  title={Clarity-2021 challenges: Machine learning challenges for advancing hearing aid processing},
  author={Graetzer, SN and Barker, Jon and Cox, Trevor J and Akeroyd, Michael and Culling, John F and Naylor, Graham and Porter, Eszter and Viveros Munoz, R and others},
  booktitle={INTERSPEECH},
  volume={2},
  pages={686--690},
  year={2021},
  organization={International Speech Communication Association (ISCA)}
}