RuntimeError: Calculated padded input size per channel: (0). Kernel size: (1). Kernel size can't be greater than actual input size

[a122760]

和#82问题一样，有处理吗

[cfso2475]

I tried to change this in Spark-TTS\cli\SparkTTS.py and seems can greatly reduce the occurance of this problem.
generated_ids = self.model.generate(
**model_inputs,
max_new_tokens=3000,
min_new_tokens=5, # add this line
do_sample=True,
top_k=top_k,
top_p=top_p,
temperature=temperature,
)