Update README.md

README.md

@@ -81,7 +81,11 @@ input_dic['spectrogram'] = torch.from_numpy(lms).unsqueez(dim=0).to(device)
 model.eval()
 gen_ids = model.forward_test(input)
 gen_text = model.neck.tokenizer.batch_decode(gen_ids.clip(0))
-
+# 4. Post-processing
+# Due to the training data, the generated texts will have some bias and require simple post-processing.
+gen_text = gen_text.split('<s>')[-1].split('\n')[0].strip()
+gen_text = gen_text.replace(' in Chinese','')
+gen_text = gen_text.replace(' Chinese','')
 print(gen_text)
 ```