Audio Voice Model to Reflect Audience Positive Emotion for Podcasting

Abstract

This study aimed to: (1) investigate the impact of podcast hosts' voices and genders on the audience's positive emotional responses, (2) develop an audio voice model to reflect the audience's positive emotions for podcasting, (3) examine the correlation between a voice model reflecting positive emotions and the gender characteristics of podcast hosts, and (4) compare the audience's positive emotions before and after listening to a podcast. The research sample comprised 10 podcast hosts (5 male and 5 female) and 50 listeners, both recruited through voluntary sampling. The study employed a mixed-method approach, utilizing quantitative and qualitative research instruments, including: (1) Podcast Impact Evaluation Form, (2) Suitability and Acceptance Evaluation Form, (3) EEG Assessment, (4) Semi-Structured Interviews, and (5) Head-Mounted Brainwave EEG Measurement to capture real-time neural activity. Data analysis included descriptive statistics, paired t-tests, Pearson correlation, and Two-Way ANOVA.

The research findings revealed that:

1) Impact of Voices and Genders of Podcast Hosts on the Audience’s Positive Emotional Responses Analyzed how podcast hosts' vocal attributes and gender influenced the emotional responses of 50 listeners, with a majority of participants aged 18–33 years old (64%). Female hosts were preferred by 64% of listeners, whereas male hosts were selected by 26%, demonstrating a significant gender-based preference in podcast engagement. The results indicate that female hosts were more effective in evoking warmth, gratitude, and passion, while male hosts elicited stronger responses in joy, excitement, and contentment. The findings reinforce the notion that tonal and spectral variations in speech significantly influence emotional engagement, with higher pitch, rhythmic modulation, and expressive articulation favoring female hosts, whereas lower pitch, steady rhythm, and dynamic intonation benefited male hosts. EEG data further validated these responses, showing increased theta and gamma wave activity for female-hosted content, which correlated with enhanced memory encoding and emotional processing.

2) Developing a Speech Model that Reflects Audience's Positive Emotions The Positive Emotions Speech Model (AVM) consists of six key components: (2.1) Input Audio Processing (2.2) Audio Analysis (2.3) Emotion Adaptive Voice Synthesis (2.4) Creative Voice Modulation (2.5) Output Audio Generation (2.6) Evaluation and Optimization. Refinement of the model through expert and listener input. AVM received the highest suitability ratings from five experts (mean = 4.80, standard deviation = 0.42). They demonstrated how the model effectively modulates alpha, beta, and theta band activity, aligning with EEG findings that show neural engagement in emotional processing and cognitive resonance. The results highlight the importance of real-time listener feedback integration in enhancing personalized podcast experiences and optimizing voice modulation for emotional impact.

3) The Pearson correlation analysis revealed significant relationships between podcast host gender and positive emotions. Female voices elicited stronger responses for gratitude, passion, and love, while male voices were more effective in generating joy, excitement, and contentment. The highlighted correlations between vocal traits (e.g., pitch, tone, rhythm) and EEG activity, confirming that gender-specific characteristics influenced brain regions associated with emotional processing. The data supported the gender-informed design of the (AVM), emphasizing its ability to align emotional delivery with listener preferences.

4) Comparison of Positive Emotions Before and After Listening A paired t-test compared positive emotions before and after podcast listening, revealing significant increases in joy, excitement, contentment, warmth, gratitude, and passion (p < .05), while hopeful remained stable. Female listeners experienced greater emotional enhancement than males, particularly in gratitude and passion. EEG data (Table 4.26, Table 4.27, Table 4.28) showed post-listening increases in theta, beta, and gamma band activities, confirming the emotional impact of the podcast on neural processing. The findings support the effectiveness of the AVM in enhancing positive emotions, demonstrating that tailored voice models can optimize emotional resonance and listener engagement in podcasting, contributing to the development of more emotionally impactful audio content.

-