jpdiazpardo commited on
Commit
accb4e2
1 Parent(s): 7034aab

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +15 -13
app.py CHANGED
@@ -19,36 +19,38 @@ def transcribe(file, task, return_timestamps):
19
  outputs = pipe(file, batch_size=BATCH_SIZE, generate_kwargs={"task": task}, return_timestamps=True)
20
  text = outputs["text"]
21
  timestamps = outputs["chunks"]
22
-
23
  if return_timestamps==True:
24
  timestamps = [f"[{format_timestamp(chunk['timestamp'][0])} -> {format_timestamp(chunk['timestamp'][1])}] {chunk['text']}" for chunk in timestamps]
25
-
26
  else:
27
  timestamps = [f"{chunk['text']}" for chunk in timestamps]
28
-
29
- text = "\n".join(str(feature) for feature in timestamps)
30
- return text
31
 
 
 
 
32
 
33
  file_transcribe = gr.Interface(
34
  fn=transcribe,
35
  inputs=[
36
- gr.inputs.Audio(source="upload", optional=True, label="Audio file", type="filepath"),
37
- gr.inputs.Radio(["transcribe"], label="Task", default="transcribe"),
38
- gr.inputs.Checkbox(default=False, label="Return timestamps"),
39
  ],
40
- outputs="text",
41
- layout="horizontal",
42
- theme="huggingface",
43
  title="Whisper Demo: Transcribe Audio",
44
  description=(
45
  "Transcribe long-form microphone or audio inputs with the click of a button! Demo uses the"
46
  f" checkpoint [{MODEL_NAME}](https://huggingface.co/{MODEL_NAME}) and 🤗 Transformers to transcribe audio files"
47
  " of arbitrary length."
48
  ),
49
-
50
  cache_examples=True,
51
  allow_flagging="never",
 
52
  )
53
 
54
- file_transcribe.launch(enable_queue=True, debug = True)
 
 
19
  outputs = pipe(file, batch_size=BATCH_SIZE, generate_kwargs={"task": task}, return_timestamps=True)
20
  text = outputs["text"]
21
  timestamps = outputs["chunks"]
22
+
23
  if return_timestamps==True:
24
  timestamps = [f"[{format_timestamp(chunk['timestamp'][0])} -> {format_timestamp(chunk['timestamp'][1])}] {chunk['text']}" for chunk in timestamps]
25
+
26
  else:
27
  timestamps = [f"{chunk['text']}" for chunk in timestamps]
 
 
 
28
 
29
+ text = "<br>".join(str(feature) for feature in timestamps)
30
+ text = f"<h4>Transcription</h4><div style='overflow-y: scroll; height: 400px;'>{text}</div>"
31
+ return file, text
32
 
33
  file_transcribe = gr.Interface(
34
  fn=transcribe,
35
  inputs=[
36
+ gr.Audio(source="upload", label="Audio file", type="filepath"),
37
+ gr.Radio(["transcribe"], label="Task", value="transcribe"),
38
+ gr.Checkbox(value=True, label="Return timestamps"),
39
  ],
40
+ outputs= [gr.Audio(label="Processed Audio", type="filepath"),
41
+ gr.outputs.HTML("text")
42
+ ],
43
  title="Whisper Demo: Transcribe Audio",
44
  description=(
45
  "Transcribe long-form microphone or audio inputs with the click of a button! Demo uses the"
46
  f" checkpoint [{MODEL_NAME}](https://huggingface.co/{MODEL_NAME}) and 🤗 Transformers to transcribe audio files"
47
  " of arbitrary length."
48
  ),
49
+
50
  cache_examples=True,
51
  allow_flagging="never",
52
+
53
  )
54
 
55
+ file_transcribe.queue(concurrency_count=3)
56
+ file_transcribe.launch(share=True, debug = True)