Poor Bat Identification

[gilpayson]

I have now hooked up a brand new Raspberry PI 5 with 16 gb ram to an Audiomoth USB v1.2 , also new, and I am getting consistently weird identifications of Bats. Bats from all the Americas - and almost none of the ones that are here.
selected options are:

MODEL=BirdNET_GLOBAL_6K_V2.4_Model_FP16
SF_THRESH=0.005 <= I have been fiddlnig this, but the identifcation is still weird

CONFIDENCE=0.8

SENSITIVITY=1
CHANNELS=1

SAMPLING_RATE=256000
BAT_CLASSIFIER=USA-EAST
LAST_CLASSIFIER=USA-EAST
BIRD_CLASSIFIER="BIRDS" <=
SWITCH_TO_BIRD="true"
BAT_TIMER="true"
BAT_SUNTIMER="true"
BAT_DUSK="18:00"
BAT_DAWN="06:00"
NOISERED="true"
NOISE_PROF="BattyBirdNET-Analyzer/checkpoints/bats/mic-noise/audiomoth_v12.prof"
NOISE_PROF_FACTOR="0.10"
INPUT_NOISERED="true"
INPUT_SPECTROGRAM_COLOR=""
SPEC_SAMPLE_RATE=256000

any ideas on what to change to improve identification? The bird ID also gets a lot of weirdness in it.

I have another birdnetpi running for over a year now - and the identification for birds alone has been excellent

thanks -Gil