Ankush Rana commited on
Commit
88a8f61
·
1 Parent(s): 23dbd9b

change to largev3

Browse files
Files changed (1) hide show
  1. app.py +4 -2
app.py CHANGED
@@ -7,7 +7,7 @@ import torch
7
  device = "cuda" if torch.cuda.is_available() else "cpu"
8
  torch_dtype = "float32"
9
 
10
- MODEL_NAME = "Systran/faster-whisper-medium"
11
  print("Loading model ...")
12
  model = WhisperModel(MODEL_NAME, compute_type=torch_dtype)
13
  print("Loading model done.")
@@ -23,6 +23,8 @@ def transcribe(inputs):
23
  # chunk_length=30,
24
  task="transcribe",
25
  word_timestamps=False,
 
 
26
  )
27
 
28
  text = ""
@@ -47,7 +49,7 @@ with gr.Blocks(theme=theme) as demo:
47
  input = gr.Audio(sources=["upload", "microphone"], type="filepath", label="Audio")
48
 
49
  with gr.Column(scale=1):
50
- output = gr.Textbox(label="Output", lines=8)
51
 
52
  with gr.Row(variant="panel"):
53
  clear_btn = gr.Button("Clear")
 
7
  device = "cuda" if torch.cuda.is_available() else "cpu"
8
  torch_dtype = "float32"
9
 
10
+ MODEL_NAME = "Systran/faster-whisper-large-v3"
11
  print("Loading model ...")
12
  model = WhisperModel(MODEL_NAME, compute_type=torch_dtype)
13
  print("Loading model done.")
 
23
  # chunk_length=30,
24
  task="transcribe",
25
  word_timestamps=False,
26
+ repetition_penalty=1.2,
27
+ without_timestamps=True
28
  )
29
 
30
  text = ""
 
49
  input = gr.Audio(sources=["upload", "microphone"], type="filepath", label="Audio")
50
 
51
  with gr.Column(scale=1):
52
+ output = gr.Textbox(label="Output")
53
 
54
  with gr.Row(variant="panel"):
55
  clear_btn = gr.Button("Clear")