Hello,
We’re researchers at Boston University investigating how bilinguals react when they're reading an emotional story in their foreign language compared to their native language.
Our research group is interested in finding a computer software that analyses auditorial conversational speech and quantifies different types of emotion and emotional intensity.
Thank you!
Sentiment analysis is generally performed by capturing the frequency of positive or negative terms in a given corpus with the help of automated analysis software such as SentiStrength, which utilizes a lexical approach for scoring positive or negative sentiment (Thelwall et al., 2012). However, since you plan to do a sentiment analysis by sound, you could check out openSMILE, which employs audio-level descriptors (Eyben et al., 2010). You might also go through the review study by Gandhi et al. (2023) for recent germane inputs. Here are the full citations.
Eyben, F., Wöllmer, M., & Schuller, B. (2010). Opensmile: The Munich versatile and fast open-source audio feature extractor. Proceedings of the International Conference on Multimedia - MM ’10, 1459. https://doi.org/10.1145/1873951.1874246
Gandhi, A., Adhvaryu, K., Poria, S., Cambria, E., & Hussain, A. (2023). Multimodal sentiment analysis: A systematic review of history, datasets, multimodal fusion methods, applications, challenges and future directions. Information Fusion, 91, 424–444. https://doi.org/10.1016/j.inffus.2022.09.025
Thelwall, M., Buckley, K., & Paltoglou, G. (2012). Sentiment strength detection for the social web. Journal of the American Society for Information Science and Technology, 63(1), 163–173. https://doi.org/10.1002/asi.21662
Good luck,
Hi Ana. Emotion in speech is usually automatically recognized on the basis of prosody. It has been found by Scherer in 2004
https://www.researchgate.net/profile/Klaus-Scherer/research
that people from different cultures classify emotionally pronounced pseudowords in a very similar way. At the same time, the machine learning techniques are language-dependent because they are trained with corpora of emotional speech in separate languages, for now, they are far from performing like humans. Recent investigations have shown that the phonemic content is also an indicator, especially of valence and some recent language-independent systems are based on it. You may take a look at this: Liu, Z. T., Rehman, A., Wu, M., Cao, W. H., & Hao, M. (2021). Speech emotion recognition based on formant characteristics feature extraction and phoneme type convergence. Information Sciences, 563, 309-325.
To me, the problem you investigate is related to how deeply a given language is related to the conceptual and emotional levels of mentally representing what is said. Well, I guess that the mother tongue is the best transmitter, isn't it?
As Martina, I suggest you to use Praat as it accesses different (formant) frequencies in the speech (which are present in the use of vowels).
Other than that, both Python and MATLAB possess packages (scipy -> signal) and toolboxes (Signal Processing Toolbox) respectively, which can be used as well.
In MATLAB, I can recommend also the following example - https://www.mathworks.com/help/signal/ug/formant-estimation-with-lpc-coefficients.html
- Badges
- Science topic
- Similar topics
- Medicine
- Epi
- Age Groups
- Children