Voice sentiment analysis gains traction

What if a machine can tell whether a human is angry, sad, happy, or indifferent? Facial recognition research has also enabled studies on facial sentiment analysis, determining the emotion of a human by analyzing the dynamics of facial muscles. But video and image analysis take a lot more horsepower and dollars to be effective, and facial recognition applications are still mainly affordable only to research institutions, law enforcement agencies, military facilities, and large casinos (according to Hollywood).

But there’s a good chance that voice sentiment analysis becomes mainstream before its facial counterpart. After all, in many instances where we cannot see each other during a communication, voice is still the next best thing. Voice sentiment analysis is a branch of speech analytics, and some speech tech vendors may already claim their software’s sentiment analysis capabilities. However, a lot of the claim is based on searching for words or phrases from a person’s speech input (for example, looking for curse words).

Obviously this isn’t the ideal approach. As we all know, these four-letter words can also be dropped into expressing excitement and positive sentiment. You and I know clearly that “I’m gonna f–king cancel the service” and “I’m so f–king happy with this product” have completely opposite sentiments. But to a machine analyzing based on keywords — both could be flagged as negative responses.

A somewhat advanced algorithm would also take into account other words in the response. For example, scoring the “cancel” as negative and “happy” as positive following the f-word.

Even better is to factor in the voice acoustics. The pitch, speed, intensity, etc. of a vocal response could offer insight into the sentiment. Pay attention to your own voice next time when you’re happy or upset, and you may notice the difference.

The benefits of good voice sentiment analysis is apparent in the contact center environment. And there appears to be traction in this niche space…

Recently, startup Saygent received $1 million in funding to take voice sentiment analysis to the cloud. The founders/makers of Saygent also operate SayHired, an automated phone screening service with some high-profile customers like Hertz, esurance, and Great Clips. Saygent would be the new kid on the block to compete with some of the more established speech analytics vendors like Autonomy, CallMiner, Nexidia, NICE, and Verint.


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s