Model returning incorrect phonemes.
#10
by
JSnack58
- opened
To Whom It May Concern,
I love what you guys are building here, but I'm having issues using the model.
The model you provided does not return the expected phonemes from the example. Did the dataset change?
Additionally, I'd get the wrong phonemes when providing my own audio. When I said the letter "l", the model returned ['a: r']. When putting this phoneme into an online phoneme reader(https://ipa-reader.com/) it sounded nothing like what I said.
I've correctly set the sampling_rate like the warning said I should, but these errors still persist.
Am I using this model incorrectly? I'd like to know.