Table of Contents
Fetching ...

Echoes of Ideology: Toward an Audio Analysis Pipeline to Unveil Character Traits in Historical Nazi Propaganda Films

Nicolas Ruth, Manuel Burghardt

TL;DR

This paper addresses how to uncover ideological narratives in Nazi-era propaganda films by developing a three-step audio analysis pipeline that combines speaker diarization, automatic transcription with Whisper, and GPT-based psycholinguistic analysis to infer character traits from dialogue. It empirically evaluates each step, finding that diarization remains the main bottleneck (DER ~58.8 with Nemo), transcription is robust yet challenged by archaic language, and GPT-based trait extraction yields plausible character portraits aligned with scholarly literature, though context attribution can be imperfect. The work demonstrates the potential for scalable, quantitative study of ideology in historical cinema and informs methodological development for archival audio analysis and distant-viewing integration, while acknowledging ethical considerations around sensitive content. Overall, it offers a pathway to quantify character portrayal and ideological patterns across Nazi propaganda films, paving the way for larger corpora analyses and cross-modal extensions.

Abstract

This study investigates the use of computational audio analysis to examine ideological narratives in Nazi propaganda films. Employing a three-step pipeline, speaker diarization, audio transcription and psycholinguistic analysis, it reveals ideological patterns in characters. Despite current issues with speaker diarization, the methodology provides insights into character traits and propaganda narratives, suggesting scalable applications.

Echoes of Ideology: Toward an Audio Analysis Pipeline to Unveil Character Traits in Historical Nazi Propaganda Films

TL;DR

This paper addresses how to uncover ideological narratives in Nazi-era propaganda films by developing a three-step audio analysis pipeline that combines speaker diarization, automatic transcription with Whisper, and GPT-based psycholinguistic analysis to infer character traits from dialogue. It empirically evaluates each step, finding that diarization remains the main bottleneck (DER ~58.8 with Nemo), transcription is robust yet challenged by archaic language, and GPT-based trait extraction yields plausible character portraits aligned with scholarly literature, though context attribution can be imperfect. The work demonstrates the potential for scalable, quantitative study of ideology in historical cinema and informs methodological development for archival audio analysis and distant-viewing integration, while acknowledging ethical considerations around sensitive content. Overall, it offers a pathway to quantify character portrayal and ideological patterns across Nazi propaganda films, paving the way for larger corpora analyses and cross-modal extensions.

Abstract

This study investigates the use of computational audio analysis to examine ideological narratives in Nazi propaganda films. Employing a three-step pipeline, speaker diarization, audio transcription and psycholinguistic analysis, it reveals ideological patterns in characters. Despite current issues with speaker diarization, the methodology provides insights into character traits and propaganda narratives, suggesting scalable applications.
Paper Structure (8 sections, 4 figures)

This paper contains 8 sections, 4 figures.

Figures (4)

  • Figure 1: Overview of the three-step analysis workflow.
  • Figure 2: System prompt and input of the GPT-based character analysis.
  • Figure 3: Output of GPT-based character analysis for the characters “Joseph Süß Oppenheimer” and the Duke, later translated to english for the paper.
  • Figure 4: Character triangle of Julieta Merck, Johannes von Redel, and Johannes's father in the Nazi youth film “Kopf hoch, Johannes!”.