whispy commited on
Commit
bb7942d
β€’
1 Parent(s): 191ffed

Upload 3 files

Browse files
Files changed (3) hide show
  1. app.py +107 -0
  2. packages.txt +1 -0
  3. requirements.txt +4 -0
app.py ADDED
@@ -0,0 +1,107 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ import torch
2
+
3
+ import gradio as gr
4
+ import pytube as pt
5
+ from transformers import pipeline
6
+
7
+
8
+ MODEL_NAME = "whispy/whisper_italian"
9
+
10
+ device = 0 if torch.cuda.is_available() else "cpu"
11
+
12
+ summarizer = pipeline(
13
+ "summarization",
14
+ model="it5/it5-efficient-small-el32-news-summarization",
15
+ )
16
+
17
+ pipe = pipeline(
18
+ task="automatic-speech-recognition",
19
+ model=MODEL_NAME,
20
+ chunk_length_s=30,
21
+ device=device,
22
+ )
23
+
24
+ translator = pipeline("translation", model="Helsinki-NLP/opus-mt-it-en")
25
+
26
+ def transcribe(microphone, file_upload):
27
+ warn_output = ""
28
+ if (microphone is not None) and (file_upload is not None):
29
+ warn_output = (
30
+ "WARNING: You've uploaded an audio file and used the microphone. "
31
+ "The recorded file from the microphone will be used and the uploaded audio will be discarded.\n"
32
+ )
33
+
34
+ elif (microphone is None) and (file_upload is None):
35
+ return "ERROR: You have to either use the microphone or upload an audio file"
36
+
37
+ file = microphone if microphone is not None else file_upload
38
+
39
+ text = pipe(file)["text"]
40
+
41
+ return warn_output + text
42
+
43
+
44
+ def _return_yt_html_embed(yt_url):
45
+ video_id = yt_url.split("?v=")[-1]
46
+ HTML_str = (
47
+ f'<center> <iframe width="500" height="320" src="https://www.youtube.com/embed/{video_id}"> </iframe>'
48
+ " </center>"
49
+ )
50
+ return HTML_str
51
+
52
+
53
+ def yt_transcribe(yt_url):
54
+ yt = pt.YouTube(yt_url)
55
+ html_embed_str = _return_yt_html_embed(yt_url)
56
+ stream = yt.streams.filter(only_audio=True)[0]
57
+ stream.download(filename="audio.mp3")
58
+
59
+ text = pipe("audio.mp3")["text"]
60
+
61
+ summary = summarizer(text)
62
+ summary = summary[0]["summary_text"]
63
+
64
+ translate = translator(summary)
65
+ translate = translate[0]["translation_text"]
66
+
67
+ return html_embed_str, text, summary, translate
68
+
69
+ demo = gr.Blocks()
70
+
71
+ mf_transcribe = gr.Interface(
72
+ fn=transcribe,
73
+ inputs=[
74
+ gr.inputs.Audio(source="microphone", type="filepath", optional=True),
75
+ gr.inputs.Audio(source="upload", type="filepath", optional=True),
76
+ ],
77
+ outputs="text",
78
+ layout="horizontal",
79
+ theme="huggingface",
80
+ title="Whisper Demo: Transcribe Audio",
81
+ description=(
82
+ "Transcribe long-form microphone or audio inputs with the click of a button! Demo uses the the fine-tuned"
83
+ f" checkpoint [{MODEL_NAME}](https://huggingface.co/{MODEL_NAME}) and πŸ€— Transformers to transcribe audio files"
84
+ " of arbitrary length."
85
+ ),
86
+ allow_flagging="never",
87
+ )
88
+
89
+ yt_transcribe = gr.Interface(
90
+ fn=yt_transcribe,
91
+ inputs=[gr.inputs.Textbox(lines=1, placeholder="Paste the URL to a YouTube video here", label="YouTube URL")],
92
+ outputs=["html", "text", "text", "text"],
93
+ layout="horizontal",
94
+ theme="huggingface",
95
+ title="Whisper Demo: Transcribe YouTube",
96
+ description=(
97
+ "Transcribe long-form YouTube videos with the click of a button! Demo uses the the fine-tuned checkpoint:"
98
+ f" [{MODEL_NAME}](https://huggingface.co/{MODEL_NAME}) and πŸ€— Transformers to transcribe audio files of"
99
+ " arbitrary length."
100
+ ),
101
+ allow_flagging="never",
102
+ )
103
+
104
+ with demo:
105
+ gr.TabbedInterface([mf_transcribe, yt_transcribe], ["Transcribe Audio", "Transcribe YouTube"])
106
+
107
+ demo.launch(enable_queue=True)
packages.txt ADDED
@@ -0,0 +1 @@
 
 
1
+ ffmpeg
requirements.txt ADDED
@@ -0,0 +1,4 @@
 
 
 
 
 
1
+ transformers
2
+ torch
3
+ pytube
4
+ sentencepiece