An emotion-based voice interaction method, a storage medium and a terminal device, the method comprising: receiving voice information to be processed, and acquiring an intent type of the voice information (S10); when the intent type is an emotion intent, determining an emotion type of the voice information (S20); generating a voice response for the voice interaction according to the emotion type, and playing the voice response (S30). The present method can generate a corresponding voice response according to emotion intent, enabling a voice response to match the emotion intent of the user, both realizing voice interaction and achieving emotional interaction, making use more convenient for a user. The emotion type of the voice information is further determined by comparing pitch, fluctuation and tone with corresponding ranges.