opencode

opencode with model inference on cheaperinference.com has been working well for me - glm-5.3-flash is shockingly cheap (i've spent a total of a few dollars over several weeks of heavy usage), fast and capable for cyber tasks

This is cool, though for me personally my minimum quality bar for the lyrics and sync accuracy in karaoke is much higher than what's achievable with WhisperX / Force Alignment. I moved on from that approach about 2 years ago in my pursuit of high quality karaoke.

If anyone's interested in tooling to generate karaoke videos which unlocks a much higher quality bar, check out my karaoke-gen project - 2 free generation credits here: https://gen.nomadkaraoke.com, open-source code here: https://github.com/nomadkaraoke/karaoke-gen