Commit
•
7bab834
1
Parent(s):
2023ef5
Update app.py
Browse files
app.py
CHANGED
@@ -1,5 +1,4 @@
|
|
1 |
import torch
|
2 |
-
|
3 |
import gradio as gr
|
4 |
import yt_dlp as youtube_dl
|
5 |
from transformers import pipeline
|
@@ -7,6 +6,7 @@ from transformers.pipelines.audio_utils import ffmpeg_read
|
|
7 |
|
8 |
import tempfile
|
9 |
import os
|
|
|
10 |
|
11 |
MODEL_NAME = "openai/whisper-large-v3"
|
12 |
BATCH_SIZE = 8
|
@@ -22,14 +22,12 @@ pipe = pipeline(
|
|
22 |
device=device,
|
23 |
)
|
24 |
|
25 |
-
|
26 |
def transcribe(inputs, task):
|
27 |
if inputs is None:
|
28 |
raise gr.Error("No audio file submitted! Please upload or record an audio file before submitting your request.")
|
29 |
|
30 |
text = pipe(inputs, batch_size=BATCH_SIZE, generate_kwargs={"task": task}, return_timestamps=True)["text"]
|
31 |
-
return
|
32 |
-
|
33 |
|
34 |
def _return_yt_html_embed(yt_url):
|
35 |
video_id = yt_url.split("?v=")[-1]
|
@@ -70,7 +68,6 @@ def download_yt_audio(yt_url, filename):
|
|
70 |
except youtube_dl.utils.ExtractorError as err:
|
71 |
raise gr.Error(str(err))
|
72 |
|
73 |
-
|
74 |
def yt_transcribe(yt_url, task, max_filesize=75.0):
|
75 |
html_embed_str = _return_yt_html_embed(yt_url)
|
76 |
|
@@ -87,14 +84,11 @@ def yt_transcribe(yt_url, task, max_filesize=75.0):
|
|
87 |
|
88 |
return html_embed_str, text
|
89 |
|
90 |
-
|
91 |
-
demo = gr.Blocks()
|
92 |
-
|
93 |
mf_transcribe = gr.Interface(
|
94 |
fn=transcribe,
|
95 |
inputs=[
|
96 |
-
gr.
|
97 |
-
gr.
|
98 |
],
|
99 |
outputs="text",
|
100 |
layout="horizontal",
|
@@ -111,8 +105,8 @@ mf_transcribe = gr.Interface(
|
|
111 |
file_transcribe = gr.Interface(
|
112 |
fn=transcribe,
|
113 |
inputs=[
|
114 |
-
gr.
|
115 |
-
gr.
|
116 |
],
|
117 |
outputs="text",
|
118 |
layout="horizontal",
|
@@ -129,8 +123,8 @@ file_transcribe = gr.Interface(
|
|
129 |
yt_transcribe = gr.Interface(
|
130 |
fn=yt_transcribe,
|
131 |
inputs=[
|
132 |
-
gr.
|
133 |
-
gr.
|
134 |
],
|
135 |
outputs=["html", "text"],
|
136 |
layout="horizontal",
|
@@ -144,8 +138,7 @@ yt_transcribe = gr.Interface(
|
|
144 |
allow_flagging="never",
|
145 |
)
|
146 |
|
147 |
-
with demo:
|
148 |
gr.TabbedInterface([mf_transcribe, file_transcribe, yt_transcribe], ["Microphone", "Audio file", "YouTube"])
|
149 |
|
150 |
demo.launch(enable_queue=True)
|
151 |
-
|
|
|
1 |
import torch
|
|
|
2 |
import gradio as gr
|
3 |
import yt_dlp as youtube_dl
|
4 |
from transformers import pipeline
|
|
|
6 |
|
7 |
import tempfile
|
8 |
import os
|
9 |
+
import time
|
10 |
|
11 |
MODEL_NAME = "openai/whisper-large-v3"
|
12 |
BATCH_SIZE = 8
|
|
|
22 |
device=device,
|
23 |
)
|
24 |
|
|
|
25 |
def transcribe(inputs, task):
|
26 |
if inputs is None:
|
27 |
raise gr.Error("No audio file submitted! Please upload or record an audio file before submitting your request.")
|
28 |
|
29 |
text = pipe(inputs, batch_size=BATCH_SIZE, generate_kwargs={"task": task}, return_timestamps=True)["text"]
|
30 |
+
return text
|
|
|
31 |
|
32 |
def _return_yt_html_embed(yt_url):
|
33 |
video_id = yt_url.split("?v=")[-1]
|
|
|
68 |
except youtube_dl.utils.ExtractorError as err:
|
69 |
raise gr.Error(str(err))
|
70 |
|
|
|
71 |
def yt_transcribe(yt_url, task, max_filesize=75.0):
|
72 |
html_embed_str = _return_yt_html_embed(yt_url)
|
73 |
|
|
|
84 |
|
85 |
return html_embed_str, text
|
86 |
|
|
|
|
|
|
|
87 |
mf_transcribe = gr.Interface(
|
88 |
fn=transcribe,
|
89 |
inputs=[
|
90 |
+
gr.Audio(source="microphone", type="filepath", optional=True),
|
91 |
+
gr.Radio(["transcribe", "translate"], label="Task", value="transcribe"),
|
92 |
],
|
93 |
outputs="text",
|
94 |
layout="horizontal",
|
|
|
105 |
file_transcribe = gr.Interface(
|
106 |
fn=transcribe,
|
107 |
inputs=[
|
108 |
+
gr.Audio(source="upload", type="filepath", optional=True, label="Audio file"),
|
109 |
+
gr.Radio(["transcribe", "translate"], label="Task", value="transcribe"),
|
110 |
],
|
111 |
outputs="text",
|
112 |
layout="horizontal",
|
|
|
123 |
yt_transcribe = gr.Interface(
|
124 |
fn=yt_transcribe,
|
125 |
inputs=[
|
126 |
+
gr.Textbox(lines=1, placeholder="Paste the URL to a YouTube video here", label="YouTube URL"),
|
127 |
+
gr.Radio(["transcribe", "translate"], label="Task", value="transcribe")
|
128 |
],
|
129 |
outputs=["html", "text"],
|
130 |
layout="horizontal",
|
|
|
138 |
allow_flagging="never",
|
139 |
)
|
140 |
|
141 |
+
with gr.Blocks() as demo:
|
142 |
gr.TabbedInterface([mf_transcribe, file_transcribe, yt_transcribe], ["Microphone", "Audio file", "YouTube"])
|
143 |
|
144 |
demo.launch(enable_queue=True)
|
|