According to the sharer's own testing, truly mature and stable real-time voice multimodal large models that are ideal choices are almost nowhere to be found on the market, whether domestic or foreign.