Real-time acoustic detection is critical for auscultation, communication, and human-artificial intelligence (AI) interactions. However, conventional acoustic sensors are bulky, rigid, and power-hungry, limiting their applicability in wearables. Developing sound perception into textiles presents a promising yet challenging pathway for intuitive and imperceptible interfaces. Here, a multifunctional, scalable, ultrasensitive, and self-powered triboelectric acoustic textile (MTA-Textile) that enables real-time sound sensing while possessing fabric properties including light weight, flexibility, and washability is reported. It supports diverse functionalities, including real-time cardiac auscultation, remote communication, voice recognition, and voice assistants, seamlessly integrated into everyday clothing. The multilayered MTA-Textile comprises a MoS2 nanocomposite coating for charge trapping and transport and a graphite-like textile for charge storage. This synergistic architecture maximizes charge generation and retention, delivering high output (18.3 V), exceptional sensitivity (3 V Pa−1) at low-intensity, low-frequency regions (<80 dB, <250 Hz), high signal-to-noise ratio (SNR) (57.5 dB), fine resolution (1 Hz), and long-term stability (36600 s retention, >10 000 cycles). The first garments serving as textile-based stethoscopes and voice intercom systems are demonstrated. With deep learning (DL)-enabled vocal command recognition, users can engage computing systems by speaking directly to the textile. This work advances next-generation acoustic wearables for healthcare, smart clothing, and human-AI interfaces.