Yes . this will require some effort to easily know speech/voice identification, similar type of data will be collected and can train a particular model through this collected data.
but language is not just speech; how can a robot understand feelings and emotions of a baby. language consists of some hidden messages conveyed by expressuions
Mother, with her own decision, is ready to die for the sake of her child's health. But what about AI robots? If this happens, it is compulsory (without the discretion). Because we give him a plan.