SoftBank is developing 'emotion-canceling' technology that alters voices to sound calmer during calls, reducing operators' psychological burden from harassment. The AI model learns vocal characteristics of anger and modifies them in real-time.
The technology changes pitch and inflection of voices without modifying content or wording, keeping a slight element of audible anger for operators to gauge customer emotions accurately.
SoftBank's AI model was trained with a dataset of 10,000 voice samples performed by Japanese actors expressing various emotions. It aims to launch the technology by March 2026.
Voice synthesis technology advancements in recent years make SoftBank's project feasible, including voice cloning with a three-second sample by Microsoft and audio-processing technology by Adobe.
[
add
]
[
|
|
...
]