Detecting and analyzing users moods is called: Sentiment Analysis.
Google explains this as; ‘’The process of computationally identifying a categorizing opinions expressed in a piece of text, especially in order to determine whether the writer’s attitude towards a particular topic, product, etc., is positive, negative, or natural.’’
In a voice-first world, this can be used to detect users moods and how the user is feeling. However, for brands and companies, this can also be used to analyze their recorded prompts or the voice AI’s tone-of-voice and test if their persona and brand values actually align with the chosen voice characteristics and words.
But where do you begin? You need to define your categories, by simply starting with ‘’positive’’ or ‘’negative’’. Make a list of positive and negative words, you can compare what the user or VUI said and categorize them accordingly. Besides, there are also open source tools like KNIME that examine how much percent of the time the user or VUI are saying negative versus positive words.
A more detailed way of testing your VUI action or skill is to test it in a Wizard of Oz kind of way. Where someone plays the records prompts or Voice AI and writing down or letting users choose what emotions or mood it triggers.
Detecting emotions and moods is a fairly new game and the company Beyond Verbal did a great job by creating an app called Moodies. This app is a great tool that allows you to test your VUI in a more advanced way. Play a 20s piece of audio from your voice AI or use your recorded prompts and it analyzes the vocal intonations in real time. It gives you an result of how your VUI is perceived and an emotional graph with a basic scale of Somber, Embracive, and Coarontational.
In this test, I spoke with an calm, slightly deeper tone-of-voice. Introducing an international tech company.
Without any research and knowing the company’s name, background, aspirations, and values? Would you suggest this could be the right tone of voice? Let me know what you think in the comments below!
However, using these emotion detections technologies does not give you 100% accuracy. It’s always needed to analyze in different ways, with multiple tools, different people, etc. Getting the VUI’s emotional state right is great, but getting it wrong can have a very high cost.
More information about VUI voice testing or audible branding for VUI? Go to www.voicebranding.nl