Measuring Emotion in the Voice: Computational Methods for Assessing Vocal Arousal
Brian Baucom, University of Utah
Primary Topic: Research Methods/Statistics
Key Words: Emotional Arousal, Affective Computing, Speech Signal Processing
Cognitive-behavioral research, assessment, and treatment revolve around spoken language. Beyond the semantic content in the words that are spoken, another critical piece of language is the vocal information-the tone, prosody, and vocal qualities that help us know if someone is excited, scared, depressed, or angry. Historically, this information was quantified via behavioral coding, but there are now efficient, reliable, and scalable computational methods for measuring the vocal expression of emotion in speech. These computational methods are readily applied to a range of applications of interest to clinical psychologists (e.g., psychotherapy, interview-based assessments) and open new opportunities for studying emotion and emotion-related processes. Plus, the only requirement is an audio recording. This AMASS will provide an applied introduction to these methods, including discussion of what can (and cannot) be studied with vocally-encoded measures, as well as how to frame hypotheses and interpret findings. In addition to introducing the basics of speech production, I will also introduce open-source, cross-platform speech processing software, guided practice in using these programs with sample recordings, and practical advice on optimizing new recordings. No previous experience with speech signal processing or computer programming is necessary.
You will learn:
1. How to use evolutionary models of speech production to inform hypothesis generation and interpretation of findings.
2. How to use open source software packages to edit audio files and to extract measures of vocal expression of emotion.
3. How to select and use equipment to improve the quality of new audio recordings.
Recommended Readings:
Juslin, P. N., & Scherer, K. (2005). Vocal expression of affect. In J. Harrigan, R. Rosenthal, & K. R. Scherer (Eds.), The new handbook of methods in nonverbal behavioral research (pp. 65-136). New York, NY: Oxford University Press. Owren, M. J., & Bachorowski, J. (2007). Measuring emotion-related vocal acoustics. In J. A. Coan, & J. J. B. Allen (Eds.), Handbook of emotion elicitation and assessment (pp. 239-266). Oxford: Oxford University Press.
Randomized controlled trials provide the empirical foundation for evidence-based behavioral practices. However, some behavioral trials turn out to be much more influential than others. What do high-impact behavioral trials have that their lower-impact cousins lack? And what sorts of studies and programmatic efforts are needed to pave the way for high-impact trials? This AMASS will focus on strategies for making behavioral RCTs as rigorous and clinically relevant as possible and for increasing their public health significance. It will address specific methodological challenges such as the selection of control groups and primary outcome measures, and describe two new conceptual frameworks that can guide the progression of clinical research efforts from treatment development and preliminary studies all the way to major, high-impact behavioral RCTs. We will discuss the advantages and disadvantages of large, simple behavioral trials relative to more complex (and, in some cases, excessively complex) RCTs. It will also emphasize research that builds bridges between behavioral, psychosocial, or psychiatric targets of intervention such as depression, PTSD, or physical inactivity, and the health-related targets of behavioral medicine interventions.
You will learn:
1. To understand the essential elements of randomized behavioral clinical trials.
2. To be able to use new models of translational research to plan and design high-impact behavioral RCTs.
3. To be able to apply principles of efficient design to maximizes the chances of success of randomized behavioral clinical trials.
Recommended Readings:
Czajkowski, S.M., Powell, L.H., Adler, N., et al. (2015). From ideas to efficacy: The ORBIT model for developing behavioral treatments for chronic diseases. Health Psychology, Feb 2. [Epub ahead of print]. Freedland, K.E., Mohr, D.C., Davidson, K.W., & Schwartz, J.E. (2011). Usual and unusual care: Existing practice control groups in randomized controlled trials of behavioral interventions. Psychosomatic Medicine, 73(4), 323-335. Gordon, D, Taddei-Peters, W, Mascette, A, Antman, M, Kaufmann, P.G., & Lauer, M.S. (2013). Publication of trials funded by the National Heart, Lung, and Blood Institute. New England Journal of Medicine, 369(20), 1926-1934.
Open All
