Search papers, labs, and topics across Lattice.
1
0
3
Surprisingly, end-to-end TTS models trained on reversed speech and text outperform those trained on conventional forward data in terms of fidelity, intelligibility, and naturalness.
