rolczynski / Automatic-Speech-Recognition

🎧 Automatic Speech Recognition: DeepSpeech & Seq2Seq (TensorFlow)
GNU Affero General Public License v3.0
222 stars 62 forks source link

Problem when training #38

Open HadiAbdullah opened 3 years ago

HadiAbdullah commented 3 years ago

Hi! When training the model, I am getting negative val_loss. Also when calling the predict function, the model always outputs empty strings. Tried different datasets. But the problem is persisting. Any help would be appreciated. Thanks!