Update app.py
Browse files
app.py
CHANGED
@@ -1,4 +1,4 @@
|
|
1 |
-
import
|
2 |
import os
|
3 |
import tempfile
|
4 |
import requests
|
@@ -8,7 +8,7 @@ import matplotlib.pyplot as plt
|
|
8 |
import torchaudio
|
9 |
import torch
|
10 |
|
11 |
-
#
|
12 |
try:
|
13 |
from speechbrain.inference import EncoderClassifier
|
14 |
speechbrain_classifier = EncoderClassifier.from_hparams(
|
@@ -17,10 +17,9 @@ try:
|
|
17 |
)
|
18 |
SPEECHBRAIN_LOADED = True
|
19 |
except Exception as e:
|
20 |
-
|
21 |
SPEECHBRAIN_LOADED = False
|
22 |
|
23 |
-
# --- Accent Analyzer Class ---
|
24 |
class AccentAnalyzer:
|
25 |
def __init__(self):
|
26 |
self.accent_profiles = {
|
@@ -91,10 +90,9 @@ class AccentAnalyzer:
|
|
91 |
"all_scores": scores
|
92 |
}
|
93 |
except Exception as e:
|
94 |
-
|
95 |
return self._simulate_accent_classification(audio_path)
|
96 |
|
97 |
-
# --- Download & Extract Audio ---
|
98 |
def download_and_extract_audio(url):
|
99 |
temp_dir = tempfile.mkdtemp()
|
100 |
video_path = os.path.join(temp_dir, "video.mp4")
|
@@ -102,7 +100,7 @@ def download_and_extract_audio(url):
|
|
102 |
|
103 |
if "youtube.com" in url or "youtu.be" in url:
|
104 |
from pytubefix import YouTube
|
105 |
-
yt = YouTube(url,
|
106 |
stream = yt.streams.filter(progressive=True, file_extension='mp4').first()
|
107 |
stream.download(output_path=temp_dir, filename="video.mp4")
|
108 |
else:
|
@@ -112,41 +110,33 @@ def download_and_extract_audio(url):
|
|
112 |
for chunk in r.iter_content(chunk_size=8192):
|
113 |
f.write(chunk)
|
114 |
|
115 |
-
# Extract audio using ffmpeg
|
116 |
subprocess.run([
|
117 |
"ffmpeg", "-i", video_path, "-ar", "16000", "-ac", "1", "-f", "wav", audio_path, "-y"
|
118 |
], stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL)
|
119 |
|
120 |
return audio_path
|
121 |
|
122 |
-
#
|
123 |
-
|
124 |
-
|
125 |
-
return "Please enter a URL.", plt.figure()
|
126 |
-
try:
|
127 |
-
audio_path = download_and_extract_audio(url)
|
128 |
-
analyzer = AccentAnalyzer()
|
129 |
-
results = analyzer.analyze_accent(audio_path)
|
130 |
-
|
131 |
-
labels, values = zip(*results["all_scores"].items())
|
132 |
-
fig, ax = plt.subplots()
|
133 |
-
ax.bar(labels, values)
|
134 |
-
ax.set_ylabel('Confidence (%)')
|
135 |
-
ax.set_title('Accent/Language Confidence')
|
136 |
-
plt.xticks(rotation=45)
|
137 |
-
plt.tight_layout()
|
138 |
|
139 |
-
|
140 |
-
|
141 |
-
|
|
|
|
|
|
|
|
|
|
|
142 |
|
143 |
-
|
144 |
-
iface = gr.Interface(
|
145 |
-
fn=analyze_from_url_gradio,
|
146 |
-
inputs=gr.Textbox(label="Enter Public Video URL (YouTube or MP4)"),
|
147 |
-
outputs=[gr.Textbox(label="Result"), gr.Plot(label="Confidence Plot")],
|
148 |
-
title="English Accent or Language Analyzer",
|
149 |
-
description="Paste a public video URL. The system will detect the accent or language spoken using SpeechBrain or simulation."
|
150 |
-
)
|
151 |
|
152 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import streamlit as st
|
2 |
import os
|
3 |
import tempfile
|
4 |
import requests
|
|
|
8 |
import torchaudio
|
9 |
import torch
|
10 |
|
11 |
+
# Load SpeechBrain
|
12 |
try:
|
13 |
from speechbrain.inference import EncoderClassifier
|
14 |
speechbrain_classifier = EncoderClassifier.from_hparams(
|
|
|
17 |
)
|
18 |
SPEECHBRAIN_LOADED = True
|
19 |
except Exception as e:
|
20 |
+
st.warning(f"Error loading SpeechBrain model: {e}. Running in simulation mode.")
|
21 |
SPEECHBRAIN_LOADED = False
|
22 |
|
|
|
23 |
class AccentAnalyzer:
|
24 |
def __init__(self):
|
25 |
self.accent_profiles = {
|
|
|
90 |
"all_scores": scores
|
91 |
}
|
92 |
except Exception as e:
|
93 |
+
st.warning(f"Fallback to simulation: {e}")
|
94 |
return self._simulate_accent_classification(audio_path)
|
95 |
|
|
|
96 |
def download_and_extract_audio(url):
|
97 |
temp_dir = tempfile.mkdtemp()
|
98 |
video_path = os.path.join(temp_dir, "video.mp4")
|
|
|
100 |
|
101 |
if "youtube.com" in url or "youtu.be" in url:
|
102 |
from pytubefix import YouTube
|
103 |
+
yt = YouTube(url, use_po_token=True, client="WEB")
|
104 |
stream = yt.streams.filter(progressive=True, file_extension='mp4').first()
|
105 |
stream.download(output_path=temp_dir, filename="video.mp4")
|
106 |
else:
|
|
|
110 |
for chunk in r.iter_content(chunk_size=8192):
|
111 |
f.write(chunk)
|
112 |
|
|
|
113 |
subprocess.run([
|
114 |
"ffmpeg", "-i", video_path, "-ar", "16000", "-ac", "1", "-f", "wav", audio_path, "-y"
|
115 |
], stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL)
|
116 |
|
117 |
return audio_path
|
118 |
|
119 |
+
# Streamlit UI
|
120 |
+
st.title("English Accent or Language Analyzer")
|
121 |
+
url = st.text_input("Enter Public Video URL (YouTube or MP4)")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
122 |
|
123 |
+
if st.button("Analyze"):
|
124 |
+
if not url:
|
125 |
+
st.error("Please enter a URL.")
|
126 |
+
else:
|
127 |
+
try:
|
128 |
+
audio_path = download_and_extract_audio(url)
|
129 |
+
analyzer = AccentAnalyzer()
|
130 |
+
results = analyzer.analyze_accent(audio_path)
|
131 |
|
132 |
+
st.markdown(results["explanation"])
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
133 |
|
134 |
+
labels, values = zip(*results["all_scores"].items())
|
135 |
+
fig, ax = plt.subplots()
|
136 |
+
ax.bar(labels, values)
|
137 |
+
ax.set_ylabel('Confidence (%)')
|
138 |
+
ax.set_title('Accent/Language Confidence')
|
139 |
+
plt.xticks(rotation=45)
|
140 |
+
st.pyplot(fig)
|
141 |
+
except Exception as e:
|
142 |
+
st.error(f"Error: {e}")
|