Run T3 inference to generate speech tokens
Description
Run T3 inference to generate speech tokens
Usage
t3_inference(
model,
cond,
text_tokens,
max_new_tokens = 1000,
temperature = 0.8,
cfg_weight = 0.5,
top_p = 0.95,
min_p = 0.05,
repetition_penalty = 1.2
)
Arguments
model: T3 modelcond: T3 conditioningtext_tokens: Tokenized text (tensor)max_new_tokens: Maximum speech tokens to generatetemperature: Sampling temperaturecfg_weight: Classifier-free guidance weighttop_p: Nucleus sampling thresholdmin_p: Minimum probability thresholdrepetition_penalty: Repetition penalty
Value
Generated speech tokens