I’ve been trying to fine tune the latest tacotron 2 model (no graves attention) from the released models page with a LJSpeech-like dataset but as i’m training it, it has a recurring error while generating certain sentences, here’s an example:
!! Error creating Test Sentence - 3
Traceback (most recent call last):
File “train.py”, line 503, in evaluate
ap.save_wav(wav, file_path)
File “/media/ssd/Projects/TTS-master/tts_namespace/TTS/utils/audio.py”, line 57, in save_wav
wav_norm = wav * (32767 / max(0.01, np.max(np.abs(wav))))
File “/media/ssd/mozillaTTS/lib/python3.6/site-packages/numpy/core/fromnumeric.py”, line 2320, in amax
out=out, **kwargs)
File “/media/ssd/mozillaTTS/lib/python3.6/site-packages/numpy/core/_methods.py”, line 26, in _amax
return umr_maximum(a, axis, None, out, keepdims)
ValueError: zero-size array to reduction operation maximum which has no identity
I unfortunately have zero knowledge in machine learning so i do not know how to debug this.