I actually want to do a follow up on the previous post before I sign off for the day! And that's answering the question... what are my other options for VO if I can't use the native audio generation for a video model like WAN/Seedance/etc! Elevenlabs just came out with their v4 model which is VERY interesting and I think does some great work when it comes to providing VO for advertisements, professional briefs, or even audiobook production. I made another comparison video using ElevenLabs' built in video-generation-for-lipsync tool so it's not going to be AS EXPRESSIVE as seedream or another purpose-built video generation tool but it's still good to look at something while you're comparing. The first take is fully v4 using emotional tags like [excited] or [conversational]. The second take is using v2 and my own recorded voice over using their voice-changing support (which is only available for v2 multi-language/english. There's a use case for everyone, in my opinion the second take sounds a bit more natural, even though v4 is a solid all-rounder.

The first one sounds too excited and artificial whereas the second is natural and clear.