Analysis of the efficacy and reliability of the Moodies app for detecting emotions through speech: Does it actually work?
[EN] Apps are software programs that enable users to optimise their resources in different areas. Recent years have seen a huge increase in the number of apps, whose use has spread in step with their perceived efficacy and reliability. This research focused on the Moodies app, designed for the voice...
| Authors: | ; ; ; |
|---|---|
| Format: | Article |
| Status: | Published version |
| Publication date: | 2020 |
| Country: | España |
| Institution: | Universidad de Salamanca (USAL) |
| Repository: | GREDOS. Repositorio Institucional de la Universidad de Salamanca |
| OAI Identifier: | oai:gredos.usal.es:10366/167318 |
| Online Access: | http://hdl.handle.net/10366/167318 |
| Access Level: | Open access |
| Keyword: | App Emotions LIWC Moodies Emotion recognition Prosody 61 Psicología |
| _version_ | 1878736371451953152 |
|---|---|
| author | Arana Martínez, José María |
| author2 | Gordillo León, Fernando Darias Plasencia, Jeannete Mestas Hernández, Lilia |
| author2_role | author author author |
| author_browse | Arana Martínez, José María Darias Plasencia, Jeannete Gordillo León, Fernando Mestas Hernández, Lilia |
| author_facet | Arana Martínez, José María Gordillo León, Fernando Darias Plasencia, Jeannete Mestas Hernández, Lilia |
| author_role | author |
| collection | GREDOS. Repositorio Institucional de la Universidad de Salamanca |
| description | [EN] Apps are software programs that enable users to optimise their resources in different areas. Recent years have seen a huge increase in the number of apps, whose use has spread in step with their perceived efficacy and reliability. This research focused on the Moodies app, designed for the voice detection of the speaker's emotions. Yet does it actually gauge emotions, and does it do so consistently over time? Our study therefore used this app to analyse the soundtracks of 34 scenes from different films in four languages, and the output Moodies provided was recorded in a brief text in English, which was processed using the tool Linguistic Inquiry and Word Count (LIWC). The same procedure was then repeated for a second measure. The analysis of the correspondence between the results obtained with Moodies and the interpretation made by LIWC considered the variables Emotion, prompted by scenes in films (disgust, happiness, anger, fear, tenderness, and sadness), Language (English, Spanish, Italian, and French), and the time of the measurement (Listening 1 and 2); an analysis was also conducted of reliability and concurrent criterion validity. The results show that Moodies correctly analyses emotions in dimensional terms (positive vs negative emotion), but not so in categorical terms, as it has difficulties in distinguishing between the emotions of anger and sadness and those of fear and disgust. In terms of reliability, there was a good correlation between listenings (r's Pearson correlation coefficient = .977), albeit with differences in the percentage of words detected (Listening 1 - Listening 2), which ranged between 0.00 and 22.06 (absolute value). It was also noted that language is not a significant variable, although it identifies a higher percentage of emotion words in scenes of fear in Spanish than in any other language. Based on these data as a whole, it may be concluded that Moodies classifies emotion in a more general way than expected and desired. |
| eu_rights_str_mv | openAccess |
| format | article |
| id | ES_8ddaef6ec77be9134c564c3e6067e1f4 |
| institution | Universidad de Salamanca (USAL) |
| instname_str | Universidad de Salamanca (USAL) |
| network_acronym_str | ES |
| network_name_str | España |
| oai_identifier_str | oai:gredos.usal.es:10366/167318 |
| publishDate | 2020 |
| publishDateSort | 2020 |
| publisher | Elsevier |
| publisherStr | Elsevier |
| reponame_str | GREDOS. Repositorio Institucional de la Universidad de Salamanca |
| spelling | Analysis of the efficacy and reliability of the Moodies app for detecting emotions through speech: Does it actually work? Arana Martínez, José María Gordillo León, Fernando Darias Plasencia, Jeannete Mestas Hernández, Lilia App Emotions LIWC Moodies Emotion recognition Prosody 61 Psicología [EN] Apps are software programs that enable users to optimise their resources in different areas. Recent years have seen a huge increase in the number of apps, whose use has spread in step with their perceived efficacy and reliability. This research focused on the Moodies app, designed for the voice detection of the speaker's emotions. Yet does it actually gauge emotions, and does it do so consistently over time? Our study therefore used this app to analyse the soundtracks of 34 scenes from different films in four languages, and the output Moodies provided was recorded in a brief text in English, which was processed using the tool Linguistic Inquiry and Word Count (LIWC). The same procedure was then repeated for a second measure. The analysis of the correspondence between the results obtained with Moodies and the interpretation made by LIWC considered the variables Emotion, prompted by scenes in films (disgust, happiness, anger, fear, tenderness, and sadness), Language (English, Spanish, Italian, and French), and the time of the measurement (Listening 1 and 2); an analysis was also conducted of reliability and concurrent criterion validity. The results show that Moodies correctly analyses emotions in dimensional terms (positive vs negative emotion), but not so in categorical terms, as it has difficulties in distinguishing between the emotions of anger and sadness and those of fear and disgust. In terms of reliability, there was a good correlation between listenings (r's Pearson correlation coefficient = .977), albeit with differences in the percentage of words detected (Listening 1 - Listening 2), which ranged between 0.00 and 22.06 (absolute value). It was also noted that language is not a significant variable, although it identifies a higher percentage of emotion words in scenes of fear in Spanish than in any other language. Based on these data as a whole, it may be concluded that Moodies classifies emotion in a more general way than expected and desired. Elsevier http://hdl.handle.net/10366/167318 |
| spellingShingle | Analysis of the efficacy and reliability of the Moodies app for detecting emotions through speech: Does it actually work? Arana Martínez, José María App Emotions LIWC Moodies Emotion recognition Prosody 61 Psicología |
| status_str | publishedVersion |
| title | Analysis of the efficacy and reliability of the Moodies app for detecting emotions through speech: Does it actually work? |
| title_full | Analysis of the efficacy and reliability of the Moodies app for detecting emotions through speech: Does it actually work? |
| title_fullStr | Analysis of the efficacy and reliability of the Moodies app for detecting emotions through speech: Does it actually work? |
| title_full_unstemmed | Analysis of the efficacy and reliability of the Moodies app for detecting emotions through speech: Does it actually work? |
| title_short | Analysis of the efficacy and reliability of the Moodies app for detecting emotions through speech: Does it actually work? |
| title_sort | Analysis of the efficacy and reliability of the Moodies app for detecting emotions through speech: Does it actually work? |
| topic | App Emotions LIWC Moodies Emotion recognition Prosody 61 Psicología |
| topic_facet | App Emotions LIWC Moodies Emotion recognition Prosody 61 Psicología |
| url | http://hdl.handle.net/10366/167318 |