Using voice analysis to reveal your survey’s hidden emotional data

Voice and video recordings are often used in Online and CAPI surveys to extract deeper responses. But what respondents say and do in these recordings doesn’t always tell you what they really think. You can, however, use voice analysis software to reveal the true emotions behind spoken words.  

Politeness isn’t always a virtue

Social norms mean people tend to behave politely when answering survey questions. And busy lifestyles mean they may not have fully thought through their responses.

Considering that 95% of purchasing decisions are subconscious1, and 38% of vocal communication comes from tone of voice, rather than words themselves2, it becomes clear that relying on words alone for market research can miss a lot of essential details. Consequentially, most researchers aren’t getting the information they need to understand what really matters to people and drives their behavior.

Unlocking emotional content

In our continual quest to keep Nfield at the forefront of professional market research, we came across a voice analysis solution called Phebi and decided to put it to the test.

Phebi analyzes speech and breaks it down into what has been said (sentiment) and how it was said (emotion). It does this by checking voice characteristics, such as pitch, rate and tone, every 3 seconds to detect and quantify the underlying emotions. Phebi’s methodology is based on well-accepted social science research and validated via 38,000+ hours of testing.

Phebi’s emotion analysis examines many factors to determine levels of five underlying emotions:

Having put it through its paces, we concluded that Phebi is a useful complement to Nfield Online and CAPI surveys instantly delivering accurate, quantified analysis of people’s nonconscious responses that goes beyond the words they say.

We were also impressed by Phebi’s auto transcription and translation function. In CAPI interviews, we found it could differentiate between interviewers’ and respondents’ voices.

Phebi features include:

  • See the emotional landscape with a one-click set of summary reports
  • View items of interest in rank order by positive emotional response
  • Get summaries that reveal the most intense emotion by topic and more
  • Understand respondents by listening to them and seeing emotions
  • Easily manage and search within audio and video responses. Uncover ‘micro-moments’ of peak emotion in longer answers. Tell stories and highlight key findings by dragging and dropping to create show reels.
  • Work cross-border with local language voice capture and translation
  • Instant, accurate, affordable transcription & translation
  • Exchange data bi-directionally via CSV or SPSS files, export via PDFs

Connecting Nfield & Phebi

Phebi can easily be used to analyze Nfield voice recordings via a simple connection.

Recordings from Nfield Online surveys can be automatically uploaded to Phebi’s server. For Nfield CAPI surveys, you should continue using the built-in Nfield recorder and upload all your audio files to Phebi in batch.

After feeding your audio recordings into Phebi, you’ll immediately be able to see various reports, such as the ones shown below, in the Phebi portal.

View concepts, etc. in rank order by net positive emotional response
Automatically generated Data Facts reveal peak emotions by topic
Show each respondent word and visualize in-the-moment emotion

If you are interested in using Phebi’s software, please contact them directly at for a live demonstration and answers to all your questions about their product.

Note: Phebi’s servers are located in the USA, Europe (Germany) and Australia.

1 G. Haltman
2 A. Mehrabian, UCLA