Search papers, labs, and topics across Lattice.
2
0
5
0
FilmBench reveals that leading video generation models fall short of cinematic standards, scoring significantly lower than on traditional benchmarks, which could redefine how we evaluate AI-generated video quality.
Taiwanese Mandarin TTS systems can achieve a 63.9% reduction in word error rates by using a context-adapted tokenizer and language model.
