Skip to content

Latest commit

 

History

History
365 lines (339 loc) · 17.8 KB

index.md

File metadata and controls

365 lines (339 loc) · 17.8 KB
layout
default
<script src="https://cdn.rawgit.com/download/polymer-cdn/1.5.0/lib/webcomponentsjs/webcomponents-lite.min.js"></script> <script src="https://code.jquery.com/jquery-3.2.1.min.js" integrity="sha256-hwg4gsxgFZhOsEEamdOYGBf13FyQuiTwlAQgxVSNgt4=" crossorigin="anonymous"></script> <script src="thirdparty/trackswitch-js/trackswitch.min.js"></script> <script type="text/javascript"> var settings = { onlyradiosolo: true, repeat: true, };
jQuery(document).ready(function() {
    jQuery(".player").trackSwitch(settings); 
});
</script>

Blind Identification of Binaural Room Impulse Responses from Smart Glasses

This page provides binaural audio examples to accompany the manuscript

Thomas Deppisch, Nils Meyer-Kahlen, Sebastia Amengual Gari, "Blind Identification of Binaural Room Impulse Responses from Smart Glasses", IEEE/ACM Transactions on Audio, Speech, and Language Processing, vol. 32, pp. 4052-4065, 2024, doi: 10.1109/TASLP.2024.3454964.

Listening Experiment

The proposed method was evaluated in a listening experiment where participants had to answer which of the test samples was recorded in the same room as the reference. The order of stimuli and trials was randomized, and the stimuli were only referred to as Reference, A, and B. For more details please see Sec. VI in the manuscript. Below you find examples of stimuli from a few trials. All stimuli of all trials are available at the GitHub repository.

The following examples compare the reference to an estimate from the same room and to a measurement from a different room.



The following examples compare the reference to an estimate from the same room and to a measurement from the same room.



The following examples compare the reference to a measurement from the same room and to a measurement from a different room.



Audio Examples from Non-Ideal Estimation Conditions

The following binaural audio examples showcase renderings under non-ideal estimation conditions (see Sec. V and especially V-F in the manuscript).

Estimation with additive babble noise of varying signal-to-noise ratio (SNR):



Estimation with inaccurate DOA (varying DOA offset):



Estimation with directional interference of varying signal-to-interference ratio (SIR):



Ringing and Resynthesis

The following binaural audio example showcases ringing that occurs in a raw BRIR estimate and suppression of the ringing using the proposed resynthesis method. See Sec. IV in the manuscript for details. Note that we deliberately chose an example where strong ringing occured.