Fine tuning of the llm and changing or cloning of desired voice
#13
by
Samexplorer
- opened
Understood that the entire components were weaved into each other, and the intermediate components pruned to reduce latency.
How can I fine tune the model or provide guiding rails for the LLM for example using RAG and other methods that improve the quality in specialized areas?
Again, How can I choose or clone a voice style for speech synthesis?