Recently, spoken language models (SLMs) have been highlighted as next-generation technology that surpasses the limitations of text-based language models by learning human speech without text to understand and generate linguistic and non-linguistic information.