fargerm commited on
Commit
58a8314
1 Parent(s): 017af4b

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +13 -0
app.py CHANGED
@@ -13,6 +13,8 @@ tokenizer = MarianTokenizer.from_pretrained(model_name)
13
  tts_model_name = "microsoft/speecht5_tts"
14
  tts_model = SpeechT5ForTextToSpeech.from_pretrained(tts_model_name)
15
  processor = SpeechT5Processor.from_pretrained(tts_model_name)
 
 
16
  speaker_embeddings = torch.tensor(load_dataset("Matthijs/cmu-arctic-xvectors", split="validation")["xvector"][0]).unsqueeze(0)
17
 
18
  # Function to translate text
@@ -30,6 +32,17 @@ def synthesize_speech(text, target_lang):
30
  # Save the speech to a file
31
  output_path = "output.wav"
32
  sf.write(output_path, speech.numpy(), samplerate=16000)
 
 
 
 
 
 
 
 
 
 
 
33
 
34
  return output_path
35
 
 
13
  tts_model_name = "microsoft/speecht5_tts"
14
  tts_model = SpeechT5ForTextToSpeech.from_pretrained(tts_model_name)
15
  processor = SpeechT5Processor.from_pretrained(tts_model_name)
16
+
17
+ # Load speaker embeddings
18
  speaker_embeddings = torch.tensor(load_dataset("Matthijs/cmu-arctic-xvectors", split="validation")["xvector"][0]).unsqueeze(0)
19
 
20
  # Function to translate text
 
32
  # Save the speech to a file
33
  output_path = "output.wav"
34
  sf.write(output_path, speech.numpy(), samplerate=16000)
35
+
36
+ # Check if the audio file was generated correctly
37
+ try:
38
+ with open(output_path, 'rb') as f:
39
+ audio_data = f.read()
40
+ if not audio_data:
41
+ st.error("Error: The audio file is empty.")
42
+ else:
43
+ st.success("Audio generated successfully.")
44
+ except Exception as e:
45
+ st.error(f"Error reading the audio file: {e}")
46
 
47
  return output_path
48