Ltx2 Talking and Singing APP

This is an app that uses LTX2’s lip-sync to generate talking videos of about 10 seconds. The model is still under development, so even small mismatches cause it to fail immediately — but it’s relatively stable compared to others, so I’m releasing it anyway. If you follow the conditions below, the success rate improves significantly.

1. Use a medium-shot image: from the waist up, with the face and shoulders clearly visible. If the image is too far away or shows the entire upper body from a distance, the failure rate increases dramatically. Think of a composition similar to a résumé photo.

2. Audio is currently limited to 10 seconds. If you want to try longer audio, check the LTX2 section in my workflow.


Audio is currently limited to 10 seconds. If you want to try longer audio, check the LTX2 section in my workflow.