citation_keywords: machine learning (ML); emotion classification; Audio emotion recognition; Nerual networks; Speech signal features citation_publication_date: 2024/03/20 theme-color: #0C4DED twitter:card: summary_large_image citation_title: Implementing machine learning techniques for continuous emotion prediction from uniformly segmented voice recordings citation_author_institution: Center for Lifespan Psychology, Max Planck Institute for Human Development, Germany keywords: machine learning (ML),emotion classification,Audio emotion recognition,Nerual networks,Speech signal features citation_publisher: Frontiers apple-mobile-web-app-title: process.env.SPACE_NAME | Articles citation_journal_title: Frontiers in Psychology description: IntroductionEmotional recognition from audio recordings is a rapidly advancing field, with significant implications for artificial intelligence and human-com... title: Frontiers | Implementing machine learning techniques for continuous emotion prediction from uniformly segmented voice recordings type: article citation_online_date: 2024/02/09 citation_issn: 1664-1078 dc:title: Frontiers | Implementing machine learning techniques for continuous emotion prediction from uniformly segmented voice recordings citation_language: English Content-Encoding: UTF-8 fb:admins: 1841006843 citation_pdf_url: https://www.frontiersin.org/journals/psychology/articles/10.3389/fpsyg.2024.1300996/pdf Content-Type: text/html; charset=UTF-8 image: https://www.frontiersin.org/files/MyHome%20Article%20Library/1300996/1300996_Thumb_400.jpg X-Parsed-By: org.apache.tika.parser.DefaultParser citation_journal_abbrev: Front. Psychol. citation_abstract: Introduction

Emotional recognition from audio recordings is a rapidly advancing field, with significant implications for artificial intelligence and human-computer interaction. This study introduces a novel method for detecting emotions from short, 1.5 s audio samples, aiming to improve accuracy and efficiency in emotion recognition technologies.

Methods

We utilized 1,510 unique audio samples from two databases in German and English to train our models. We extracted various features for emotion prediction, employing Deep Neural Networks (DNN) for general feature analysis, Convolutional Neural Networks (CNN) for spectrogram analysis, and a hybrid model combining both approaches (C-DNN). The study addressed challenges associated with dataset heterogeneity, language differences, and the complexities of audio sample trimming.

Results

Our models demonstrated accuracy significantly surpassing random guessing, aligning closely with human evaluative benchmarks. This indicates the effectiveness of our approach in recognizing emotional states from brief audio clips.

Discussion

Despite the challenges of integrating diverse datasets and managing short audio samples, our findings suggest considerable potential for this methodology in real-time emotion detection from continuous speech. This could contribute to improving the emotional intelligence of AI and its applications in various areas.

 citation_author: Diemerling, Hannes url: https://www.frontiersin.org/journals/psychology/articles/10.3389/fpsyg.2024.1300996/full site_name: Frontiers citation_firstpage: 1300996 viewport: width=device-width, initial-scale=1 citation_doi: 10.3389/fpsyg.2024.1300996 mobile-web-app-capable: yes dc.identifier: doi:10.3389/fpsyg.2024.1300996 citation_volume: 15 Content-Language: en