For the most realistic results, always use a clean audio file (no background noise) and manually tweak the jaw movement afterward to add "weight" to the speech.
If Rhubarb feels too technical, the add-on by CGDash (available on Blender Market, ~$10-20) offers a native feel. It has a built-in spectrogram and allows you to manually drag viseme bars onto a timeline while listening to audio. auto lip sync blender
in the 3D Viewport to open the right side panel and select the new Import Audio: Select the spoken audio file you want to sync. Map the Shapes: For the most realistic results, always use a