OpenMOSS’s MOSS-TTS-Nano-100M delivers multilingual speech synthesis, streaming output, and reference-based voice cloning without GPU hardware.