Steering latent voice representations for voice reconstruction based on a witness’s memory trace

StatusVoR
Dublin Core metadata
Alternative title
Koncepcja metody modyfikacji ukrytych reprezentacji głosu w celu jego rekonstrukcji na podstawie śladu pamięciowego świadka
Authors
Olawski, Jakub
Kuś, Filip
Dukała, Karolina
Witkowski, Marcin
Monograph
Monograph (alternative title)
Date
2026
Publisher
Journal title
Z Zagadnien Nauk Sądowych
Issue
Volume
145
Pages
Pages
45–63
DOI
ISSN
1230-7483
ISSN of series
Weblink
https://ejournals.eu/czasopismo/problems-of-forensic-sciences/artykul/steering-latent-voice-representations-for-voice-reconstruction-based-on-a-witnesss-memory-trace
Access date
2026-09-14
Abstract PL
Abstract EN
This article addresses the significant scarcity of forensic tools for generating voice samples for identification parades by proposing a novel method for voice reconstruction without a reference recording. The presented solution serves as an auditory analogue to facial composites, utilizing a hybrid architecture based on machine learning and speech synthesis (XTTS). The methodology combines an iterative algorithm for selecting a base voice candidate with a specialized neural network module that allows for the modification of interpretable acoustic parameters via latent space manipulation. Technical validation confirmed the system’s effectiveness in navigating the voice space and accurately translating physical parameters into vector representations. By shifting the identification burden from error-prone verbal descriptions to direct auditory perception, the system minimizes the verbal overshadowing effect. Consequently, the proposed prototype offers promising practical implications for forensic science, providing a technological foundation for law enforcement to conduct accurate voice lineups based solely on witness memory.
Abstract other
Keywords PL
Keywords EN
forensic acoustics
voice reconstruction
machine learning
voice lineup
auditory memory
latent space
Keywords other
Sustainable Development Goals
SDG 16. Peace, Justice and Strong Institutions
Exhibition title
Place of exhibition (institution)
Exhibition curator
URI
https://share.swps.edu.pl/handle/swps/2528
Type
License type
cc-by-nc-nd
Except as otherwise noted, this item is licensed under the Attribution-NonCommercial-NoDerivatives licence | Permitted use of copyrighted works
Funder
Time range from
Time range to
Contact person name
Related publication
Related publication
Grant/project name
Grant number
Files
pdf
Steering latent voice representations for voice reconstruction based on a witness’s memory trace.pdf(12.35 MB)
Download
Views
Views0
Acquisition Date20.10.2022
Downloads
Downloads0
Acquisition Date20.10.2022
Altmetrics©
Dimensions
Google Scholar
Google Scholar