multiple wavs for voice cloning

[Zibri]

This is my "program":

from TTS.api import TTS
tts = TTS("tts_models/multilingual/multi-dataset/xtts_v2").to("cpu")
t=open('text.txt', 'r').read().replace('\n','')
tts.tts_to_file(text=t, speaker_wav=["./speakerA.wav","speakerB.wav"], language="en", file_path="testAB.wav")

and this other program:

from TTS.api import TTS
tts = TTS("tts_models/multilingual/multi-dataset/xtts_v2").to("cpu")
t=open('text.txt', 'r').read().replace('\n','')
tts.tts_to_file(text=t, speaker_wav=["./speakerB.wav","speakerA.wav"], language="en", file_path="testBA.wav")

The two results are different. Why?
Isn't it possible to average the two so to have the same result, no matter the order?

[chen-wei-lin]

cause the speaker_wav is different ?