Update app.py

#10
by Yusin - opened
Files changed (1) hide show
  1. app.py +30 -30
app.py CHANGED
@@ -6,9 +6,9 @@ default_lang = "en"
6
 
7
  # ChatGPT
8
  from pyChatGPT import ChatGPT
9
- #import whisper
10
- #whisper_model = whisper.load_model("small")
11
- whisper = gr.Blocks.load(name="spaces/sanchit-gandhi/whisper-large-v2")
12
  import os
13
  session_token = os.environ.get('SessionToken')
14
 
@@ -46,30 +46,6 @@ def chat_hf(audio, custom_token, language):
46
  return whisper_text, gpt_response, fp.name
47
 
48
  # whisper
49
- #def translate(audio):
50
- # print("""
51
- # β€”
52
- # Sending audio to Whisper ...
53
- # β€”
54
- # """)
55
- #
56
- # audio = whisper.load_audio(audio)
57
- # audio = whisper.pad_or_trim(audio)
58
- #
59
- # mel = whisper.log_mel_spectrogram(audio).to(whisper_model.device)
60
- #
61
- # _, probs = whisper_model.detect_language(mel)
62
- #
63
- # transcript_options = whisper.DecodingOptions(task="transcribe", fp16 = False)
64
- #
65
- # transcription = whisper.decode(whisper_model, mel, transcript_options)
66
- #
67
- # print("language spoken: " + transcription.language)
68
- # print("transcript: " + transcription.text)
69
- # print("β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”")
70
- #
71
- # return transcription.text
72
-
73
  def translate(audio):
74
  print("""
75
  β€”
@@ -77,9 +53,33 @@ def translate(audio):
77
  β€”
78
  """)
79
 
80
- text_result = whisper(audio, None, "transcribe", fn_index=0)
81
- print(text_result)
82
- return text_result
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
83
 
84
 
85
  with gr.Blocks() as blocks:
 
6
 
7
  # ChatGPT
8
  from pyChatGPT import ChatGPT
9
+ import whisper
10
+ whisper_model = whisper.load_model("small")
11
+ #whisper = gr.Blocks.load(name="spaces/sanchit-gandhi/whisper-large-v2")
12
  import os
13
  session_token = os.environ.get('SessionToken')
14
 
 
46
  return whisper_text, gpt_response, fp.name
47
 
48
  # whisper
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
49
  def translate(audio):
50
  print("""
51
  β€”
 
53
  β€”
54
  """)
55
 
56
+ audio = whisper.load_audio(audio)
57
+ audio = whisper.pad_or_trim(audio)
58
+
59
+ mel = whisper.log_mel_spectrogram(audio).to(whisper_model.device)
60
+
61
+ _, probs = whisper_model.detect_language(mel)
62
+
63
+ transcript_options = whisper.DecodingOptions(task="transcribe", fp16 = False)
64
+
65
+ transcription = whisper.decode(whisper_model, mel, transcript_options)
66
+
67
+ print("language spoken: " + transcription.language)
68
+ print("transcript: " + transcription.text)
69
+ print("β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”β€”")
70
+
71
+ return transcription.text
72
+
73
+ #def translate(audio):
74
+ # print("""
75
+ # β€”
76
+ # Sending audio to Whisper ...
77
+ # β€”
78
+ # """)
79
+ #
80
+ # text_result = whisper(audio, None, "transcribe", fn_index=0)
81
+ # print(text_result)
82
+ # return text_result
83
 
84
 
85
  with gr.Blocks() as blocks: