Impact of Pathological Speech on Speaker Anonymization – A Proof of Concept (en)

* Presenting author
Day / Time: 09.03.2023, 11:00-11:40
Typ: Poster
Information: Die Poster werden durchgängig von Dienstag bis Donnerstag im Foyer auf Ebene Z ausgestellt (3.OG vor dem Plenarsaal Z), sortiert nach thematischen Zusammenhängen in der hier angegebenen Posterinsel (siehe Sitzungstitel). Das Posterforum zur angegebenen Zeit bietet die Möglichkeit, mit den Autor*innen in Diskussion zu kommen.
Abstract: Speaker anonymization is important for identity concealment in different contexts. In this work, we examine the effects of pathological speech on the anonymization process. We are differentiating between two scenarios, a pathology-preserving anonymization in the diagnostic or remote treatment environment and a pathology-obfuscating anonymization as a general approach.In our contribution, we show how different state-of-the-art approaches perform at this task. First results show that when utilizing a speaker conversion method, the pathologies are still present after anonymization. This effect is not only audible and therefore improving subjective recognition rates, but also shown in the objective evaluation. While in a remote treatment scenario for speech pathologies, this solution might still be useful, it presents a vulnerability in the anonymization. To avoid re-identification by these pathologies, we propose another approach utilizing automatic speech recognition and text based speech synthesis.


strpos(): Passing null to parameter #1 ($haystack) of type string is deprecated

/kunden/431424_13355/webseiten/subdomains/, line 35
32.              */
33.             if (headers_sent()) {
34.                 $encoding = false;
35.             } elseif (strpos($HTTP_ACCEPT_ENCODING, 'x-gzip') !== false) {
36.                 $encoding = 'x-gzip';
37.             } elseif (strpos($HTTP_ACCEPT_ENCODING, 'gzip') !== false) {
38.                 $encoding = 'gzip';