Teleperformance SE, known as the world’s largest call center operator, is rolling out an AI system aimed at moderating the accents of English-speaking Indian workers in real-time. This technology, according to the company, is expected to boost customer comprehension and satisfaction.
“When an Indian agent is on the call, it can sometimes be challenging for customers to understand,” Deputy Chief Executive Officer Thomas Mackenbrock mentioned in his conversation with Bloomberg. “[The AI can] neutralize the accent of the Indian speaker with no delay, fostering greater intimacy, enhancing customer satisfaction, and reducing the average handling time.”
The technology examines speech input and alters it to fit a designated accent while maintaining the original voice and emotion of the speaker. The system leverages speech recognition to capture the speaker’s vocal input in real-time. Using sophisticated algorithms, it transcribes spoken words into text, considering various accents and speech nuances. After transcription, the core accent translation mechanism analyzes the speaker’s accent and pronunciation traits. It then utilizes AI models trained with datasets of different accents to adjust the voice.
Throughout the conversion process, the software alters intonation, stress patterns, and phoneme pronunciation to correspond with the target accent. The system then integrates text-to-speech (TTS) technology to create the phonetic pattern and convert the text into a synthesized voice that retains the speaker’s original tone, emotion, and identity.
Teleperformance is implementing this technology with noise-cancellation features in Indian call centers, serving international client support. Its client roster includes tech giants like Apple, TikTok, and Samsung Electronics.
This technology is part of a broader Teleperformance AI investment agenda. The company is planning to allocate up to €100 million ($104 million) towards AI partnerships in 2025. Sanas, a startup based in Palo Alto, developed the accent technology following a $13 million investment from Teleperformance earlier this year.
The emergence of this software comes at a time when the call center industry is navigating challenges posed by the rise of AI chatbots. Last year, Teleperformance experienced a notable decline in share prices after Swedish fintech company Klarna Bank disclosed that its AI assistant was handling tasks equivalent to 700 full-time agents. Consequently, Teleperformance has directed its efforts on leveraging AI to complement rather than replace its workforce of 490,000 employees.
Sanas asserts that its aim is to curb “accent-based discrimination.” However, while the software might not directly replace workers, there are concerns regarding the potential repercussions on call centers in regions like the Philippines, which have established their competitive edge on the basis of high-quality English speakers.
Presently, the software supports Indian and Filipino accents. Sanas is working on developing versions for other areas, such as Latin America. Mackenbrock emphasized that despite AI’s growing presence, “the human element will remain incredibly important” in establishing connections and enriching customer experiences.
