Sarashina2.2-TTS: Tackling Kanji Polyphony in Japanese Speech Generation via Data Scaling and Targeted Data Synthesis プレプリント Lianbo Liu・Shiao Zhu・Kai Washizaki・Reo Yoneyama・Haesung Jeon・Mengjie Zhao・Yusuke Fujita・H. Z. Shi・Nao Yoshida・Gao Y・Roman Koshkin・Yukiya Hono・Yui Sudo ·2026-06-24 ·Cornell University arxiv.org