Mantella is a Skyrim and Fallout 4 mod which allows you to naturally speak to NPCs using Whisper (speech-to-text), LLMs (text generation), and xVASynth / XTTS (text-to-speech).
The character_df is passed to the tts.Synthesizer to build two dictionaries for xTTS based on 'advanced_voice_model' and 'voice_model', which are compared to the return value of xtts_get_speakers_list.
I implemented this so that when there is a need to search for 'advanced_voice_model' or 'voice_model', it will not browse the entire CSV values but only those that can be present inside the xTTS /speakers folder and /latent_speaker_folder.
Additionally, two .csv files are created in the root folder that lists the used/not used speakers across 'advanced_voice_model' and 'voice_model' based on the comparison that occurs at the start.
It needs to use the new xTTS API Mantella server 1.3 to work.
I've cleaned up the way it changes voice and removed some unnecessary calls.
I did test xVASynth this time, so it should be okay!
The character_df is passed to the tts.Synthesizer to build two dictionaries for xTTS based on 'advanced_voice_model' and 'voice_model', which are compared to the return value of xtts_get_speakers_list.
I implemented this so that when there is a need to search for 'advanced_voice_model' or 'voice_model', it will not browse the entire CSV values but only those that can be present inside the xTTS /speakers folder and /latent_speaker_folder.
Additionally, two .csv files are created in the root folder that lists the used/not used speakers across 'advanced_voice_model' and 'voice_model' based on the comparison that occurs at the start.
It needs to use the new xTTS API Mantella server 1.3 to work.
I've cleaned up the way it changes voice and removed some unnecessary calls.
I did test xVASynth this time, so it should be okay!