how do i use pregenerated conditioning latents?

[lobsterchan27]

im using the get_conditioning_latents method in api_fast. but it seems that neither the tts nor tts_with_preset methods use conditioning_latents as arguments. tts_stream accepts the passed argument but the method itself doesn't seem to do anything with it but instead opts to use random generated latents if voice_samples arent passed.

the docstring for all 3 seem to imply that passing conditioning latents is supported however it states it expects a tuple of (autoregressive_conditioning_latent, diffusion_conditioning_latent); it does not and get_condition_latents returns a single value.

am i missing something? im not too familiar with tortoise nor python

thanks!

[lukestogsdill]

I am having the same issue, please help!

[quietMachi]

I keep running into this problem, what is in the actual code doesn't seem add up with what the documentation is asking for. I assumed the code would work like intended but it's as if I was left a barely working decoy.