bark

Bark is an open-source, transformer-based text-to-audio model developed by Suno that generates highly realistic, multilingual speech alongside other audio types, including music and nonverbal sounds. Its primary use case lies in research and commercial applications requiring dynamic audio generation, with features supporting pretrained models for quick integration, low VRAM GPU compatibility, and voice consistency enhancements. Notably, Bark's capability to produce various sound outputs, including laughter and sighs, sets it apart from conventional text-to-speech models.