Spaces:

Mihaj
/

Mihaj-wav2vec2-large-xls-r-300m-ruOH-alphav

Running

Mihaj commited on 25 days ago

Commit

0f9c96a

•

1 Parent(s): 1d5a081

Update app.py

Files changed (1) hide show

app.py CHANGED Viewed

@@ -81,7 +81,7 @@ def fast_transcribe(diarise, how_diarise, translate, audio):
                       input_ids = tokenizer_tr(src_text, return_tensors="pt")
                       generated_tokens = model_tr.generate(**input_ids.to(device))
-                      trans_eng = tokenizer_tr.batch_decode(generated_tokens, skip_special_tokens=True)
                       #trans_eng = translator.translate(trans, src='ru', dest="en").text
                       print(f"TRANSLATION ENDED RESULT {trans_eng}")
                       lines.append(f"{i+1}\n{start_time_srt} --> {end_time_srt}\n[{label}] {trans}\n[{label}] {trans_eng}\n")
@@ -114,7 +114,7 @@ def fast_transcribe(diarise, how_diarise, translate, audio):
                   input_ids = tokenizer_tr(src_text, return_tensors="pt")
                   generated_tokens = model_tr.generate(**input_ids.to(device))
-                  trans_eng = tokenizer_tr.batch_decode(generated_tokens, skip_special_tokens=True)
                   #trans_eng = translator.translate(trans, src='ru', dest="en").text
                   print(f"TRANSLATION ENDED RESULT {trans_eng}")
                   lines.append(f"{i+1}\n{start_time_srt} --> {end_time_srt}\n{trans}\n{trans_eng}\n")

                       input_ids = tokenizer_tr(src_text, return_tensors="pt")
                       generated_tokens = model_tr.generate(**input_ids.to(device))
+                      trans_eng = tokenizer_tr.batch_decode(generated_tokens, skip_special_tokens=True)[0]
                       #trans_eng = translator.translate(trans, src='ru', dest="en").text
                       print(f"TRANSLATION ENDED RESULT {trans_eng}")
                       lines.append(f"{i+1}\n{start_time_srt} --> {end_time_srt}\n[{label}] {trans}\n[{label}] {trans_eng}\n")
                   input_ids = tokenizer_tr(src_text, return_tensors="pt")
                   generated_tokens = model_tr.generate(**input_ids.to(device))
+                  trans_eng = tokenizer_tr.batch_decode(generated_tokens, skip_special_tokens=True)[0]
                   #trans_eng = translator.translate(trans, src='ru', dest="en").text
                   print(f"TRANSLATION ENDED RESULT {trans_eng}")
                   lines.append(f"{i+1}\n{start_time_srt} --> {end_time_srt}\n{trans}\n{trans_eng}\n")