Search papers, labs, and topics across Lattice.
4
0
8
4
A unified music identification system can achieve robust performance across track and version identification tasks with just 10 seconds of audio input.
Current music-understanding LLMs can't tell you *when* something happens in a song, but a new benchmark and training recipe, MusTBENCH and MusT, can help them learn.
Unleashing multilingual rollouts and adaptively routing languages during policy optimization can significantly boost performance, proving that language diversity is a valuable, but often overlooked, training signal.
Woosh leapfrogs existing open models in sound effect generation, offering researchers a new high-quality foundation model and paving the way for more realistic and immersive audio experiences.
