Japanese technological giant SoftBank Corp. has developed an innovative solution to tackle the escalating issue of customer aggression in call centers. The company has introduced voting technology based on artificial intelligence (AI), capable of transforming even the most irate voices into calmer tones.
The system, known as “emotional suppression,” aims to reduce the stress experienced by call center operators who frequently encounter discontented customers. According to a press release from SoftBank here, “We are focused on developing a solution that can convert the client’s voice into a soothing conversational tone and deliver it to our employees using emotion recognition and AI-based voice processing technology.”
The company has underscored the significance of maintaining positive client relationships while ensuring the psychological well-being of its employees.
The creation of this system was prompted by a television program highlighting verbal abuse directed at call center staff. SoftBank employee Toshiyuki Nakadani was motivated to devise a solution to safeguard his colleagues from such harassment.
The technology functions in two stages. Initially, AI analyzes a voice to detect irate customers and scrutinize the speech characteristics. Subsequently, it utilizes the acoustic attributes of neutral voices to generate a more composed and natural tone.
To train this AI, over 10,000 voice data samples recorded by 10 actors articulating more than 100 phrases conveying various emotions, including anger and disappointment, were utilized. While the technology does not alter the customer’s words, it notably adjusts the intonation, making it less confrontational.
However, SoftBank assures that the system does not entirely eradicate signs of anger, enabling operators to comprehend the situation and respond appropriately.
Currently, the “Emotional suppression” system operates exclusively in Japanese. Nevertheless, the company is contemplating expanding the technology to other languages to cater to international markets.
The initial systems are projected to be rolled out by April next year, although their cost has not been determined yet.