Spaces:
Sleeping
Sleeping
Shahzad8515
commited on
Update app.py
Browse files
app.py
CHANGED
@@ -35,13 +35,10 @@ def audio_to_text(audio_file_path):
|
|
35 |
with sr.AudioFile(audio_file_path) as source:
|
36 |
audio = recognizer.record(source)
|
37 |
text = recognizer.recognize_google(audio)
|
38 |
-
print(f"Extracted Text: {text}") # Debugging line
|
39 |
return text
|
40 |
except sr.UnknownValueError:
|
41 |
-
print("Audio could not be understood") # Debugging line
|
42 |
return None
|
43 |
except sr.RequestError:
|
44 |
-
print("Request error") # Debugging line
|
45 |
return None
|
46 |
|
47 |
# Function to convert audio to WAV format
|
@@ -53,8 +50,7 @@ def convert_to_wav(audio_file_path):
|
|
53 |
wav_path = "temp_audio.wav"
|
54 |
audio.export(wav_path, format="wav")
|
55 |
return wav_path
|
56 |
-
except Exception
|
57 |
-
print(f"Error converting audio to WAV: {e}")
|
58 |
return None
|
59 |
|
60 |
# Function to extract text from a PDF file
|
@@ -65,8 +61,8 @@ def extract_text_from_pdf(pdf_file):
|
|
65 |
for page_num in range(len(pdf_document)):
|
66 |
page = pdf_document.load_page(page_num)
|
67 |
text += page.get_text()
|
68 |
-
except Exception
|
69 |
-
|
70 |
return text
|
71 |
|
72 |
# Function to embed text using a transformer model
|
@@ -76,8 +72,7 @@ def embed_text(texts, model, tokenizer):
|
|
76 |
with torch.no_grad():
|
77 |
embeddings = model(**inputs).last_hidden_state.mean(dim=1).numpy()
|
78 |
return embeddings
|
79 |
-
except Exception
|
80 |
-
print(f"Error embedding text: {e}")
|
81 |
return np.array([]) # Return empty array on error
|
82 |
|
83 |
# Function to convert text to speech
|
@@ -86,8 +81,7 @@ def text_to_speech(text, output_file):
|
|
86 |
tts = gTTS(text=text, lang='en')
|
87 |
tts.save(output_file)
|
88 |
return output_file
|
89 |
-
except Exception
|
90 |
-
print(f"Error converting text to speech: {e}")
|
91 |
return None
|
92 |
|
93 |
# Read all PDF files from the specified folder
|
@@ -98,25 +92,17 @@ for path in pdf_paths:
|
|
98 |
pdf_text = extract_text_from_pdf(path)
|
99 |
if pdf_text:
|
100 |
texts.append(pdf_text)
|
101 |
-
else:
|
102 |
-
print(f"Failed to extract text from {path}")
|
103 |
|
104 |
# Embed PDF texts and add to vector database
|
105 |
embeddings = embed_text(texts, model, tokenizer)
|
106 |
if embeddings.size > 0:
|
107 |
index.add(embeddings)
|
108 |
-
else:
|
109 |
-
print("No embeddings to add to the vector database")
|
110 |
|
111 |
def process_audio(audio_file):
|
112 |
if audio_file is None:
|
113 |
-
return "No audio file provided", None
|
114 |
-
|
115 |
-
if isinstance(audio_file, str):
|
116 |
-
audio_file_path = audio_file
|
117 |
-
else:
|
118 |
-
audio_file_path = audio_file.name
|
119 |
|
|
|
120 |
wav_path = convert_to_wav(audio_file_path)
|
121 |
if wav_path is None:
|
122 |
return "Error converting audio file to WAV format", None
|
@@ -138,15 +124,12 @@ def process_audio(audio_file):
|
|
138 |
|
139 |
if not combined_text.strip():
|
140 |
return "No relevant information found in the PDFs", None
|
|
|
141 |
prompt = (
|
142 |
-
|
143 |
-
|
144 |
-
|
145 |
-
|
146 |
-
|
147 |
-
|
148 |
-
|
149 |
-
print(f"Prompt: {prompt}") # Debugging line
|
150 |
|
151 |
chat_completion = client.chat.completions.create(
|
152 |
messages=[
|
@@ -165,30 +148,51 @@ def process_audio(audio_file):
|
|
165 |
return "Error generating speech output", None
|
166 |
|
167 |
return response, output_path
|
168 |
-
except Exception
|
169 |
-
print(f"Error in process_audio: {e}")
|
170 |
return "An error occurred while processing the audio", None
|
171 |
|
|
|
172 |
iface = gr.Interface(
|
173 |
fn=process_audio,
|
174 |
inputs=gr.Audio(type="filepath"),
|
175 |
-
outputs=[gr.Textbox(label="Advice"), gr.Audio(label="Advice Audio")],
|
176 |
-
title="BetterCrops",
|
177 |
-
description=
|
178 |
-
|
|
|
|
|
179 |
article=(
|
180 |
-
"<div style='text-align: center; color: #
|
181 |
-
"<h1>BetterCrops</h1>"
|
182 |
-
"<h3 style='font-size: 24px;'>
|
183 |
"</div>"
|
184 |
),
|
|
|
185 |
css=(
|
186 |
-
"
|
187 |
-
|
188 |
-
|
189 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
190 |
)
|
191 |
)
|
192 |
|
193 |
if __name__ == "__main__":
|
194 |
-
iface.launch()
|
|
|
35 |
with sr.AudioFile(audio_file_path) as source:
|
36 |
audio = recognizer.record(source)
|
37 |
text = recognizer.recognize_google(audio)
|
|
|
38 |
return text
|
39 |
except sr.UnknownValueError:
|
|
|
40 |
return None
|
41 |
except sr.RequestError:
|
|
|
42 |
return None
|
43 |
|
44 |
# Function to convert audio to WAV format
|
|
|
50 |
wav_path = "temp_audio.wav"
|
51 |
audio.export(wav_path, format="wav")
|
52 |
return wav_path
|
53 |
+
except Exception:
|
|
|
54 |
return None
|
55 |
|
56 |
# Function to extract text from a PDF file
|
|
|
61 |
for page_num in range(len(pdf_document)):
|
62 |
page = pdf_document.load_page(page_num)
|
63 |
text += page.get_text()
|
64 |
+
except Exception:
|
65 |
+
pass
|
66 |
return text
|
67 |
|
68 |
# Function to embed text using a transformer model
|
|
|
72 |
with torch.no_grad():
|
73 |
embeddings = model(**inputs).last_hidden_state.mean(dim=1).numpy()
|
74 |
return embeddings
|
75 |
+
except Exception:
|
|
|
76 |
return np.array([]) # Return empty array on error
|
77 |
|
78 |
# Function to convert text to speech
|
|
|
81 |
tts = gTTS(text=text, lang='en')
|
82 |
tts.save(output_file)
|
83 |
return output_file
|
84 |
+
except Exception:
|
|
|
85 |
return None
|
86 |
|
87 |
# Read all PDF files from the specified folder
|
|
|
92 |
pdf_text = extract_text_from_pdf(path)
|
93 |
if pdf_text:
|
94 |
texts.append(pdf_text)
|
|
|
|
|
95 |
|
96 |
# Embed PDF texts and add to vector database
|
97 |
embeddings = embed_text(texts, model, tokenizer)
|
98 |
if embeddings.size > 0:
|
99 |
index.add(embeddings)
|
|
|
|
|
100 |
|
101 |
def process_audio(audio_file):
|
102 |
if audio_file is None:
|
103 |
+
return "No audio file provided", None
|
|
|
|
|
|
|
|
|
|
|
104 |
|
105 |
+
audio_file_path = audio_file if isinstance(audio_file, str) else audio_file.name
|
106 |
wav_path = convert_to_wav(audio_file_path)
|
107 |
if wav_path is None:
|
108 |
return "Error converting audio file to WAV format", None
|
|
|
124 |
|
125 |
if not combined_text.strip():
|
126 |
return "No relevant information found in the PDFs", None
|
127 |
+
|
128 |
prompt = (
|
129 |
+
f"The user has asked a query related to agricultural practices: {text}. "
|
130 |
+
f"Here are relevant excerpts from the Better Crops South Asia document: {combined_text}. "
|
131 |
+
"Based on this information, please provide accurate advice related to sustainable crop management, pest control, irrigation practices, and any recommendations for improving crop yield in the South Asian region."
|
132 |
+
)
|
|
|
|
|
|
|
|
|
133 |
|
134 |
chat_completion = client.chat.completions.create(
|
135 |
messages=[
|
|
|
148 |
return "Error generating speech output", None
|
149 |
|
150 |
return response, output_path
|
151 |
+
except Exception:
|
|
|
152 |
return "An error occurred while processing the audio", None
|
153 |
|
154 |
+
# Enhanced Gradio interface customization
|
155 |
iface = gr.Interface(
|
156 |
fn=process_audio,
|
157 |
inputs=gr.Audio(type="filepath"),
|
158 |
+
outputs=[gr.Textbox(label="Advice", lines=10), gr.Audio(label="Advice Audio")],
|
159 |
+
title="🌾 BetterCrops: Agriculture Support for Farmers",
|
160 |
+
description=(
|
161 |
+
"💡 **BetterCrops** is designed to assist farmers with their crops by analyzing agricultural PDFs "
|
162 |
+
"and generating personalized audio advice based on your voice queries."
|
163 |
+
),
|
164 |
article=(
|
165 |
+
"<div style='text-align: center; color: #003f6e;'>"
|
166 |
+
"<h1 style='font-size: 36px; font-weight: bold;'>BetterCrops</h1>"
|
167 |
+
"<h3 style='font-size: 24px; font-weight: normal;'>Empowering Farmers with AI-driven Insights</h3>"
|
168 |
"</div>"
|
169 |
),
|
170 |
+
theme="grass",
|
171 |
css=(
|
172 |
+
"""
|
173 |
+
body {
|
174 |
+
background-color: #f0f5e9;
|
175 |
+
color: #2f4f2f;
|
176 |
+
font-family: 'Helvetica Neue', sans-serif;
|
177 |
+
}
|
178 |
+
h1, h3 {
|
179 |
+
color: #003f6e;
|
180 |
+
}
|
181 |
+
.gradio-container {
|
182 |
+
padding: 20px;
|
183 |
+
background: linear-gradient(135deg, #a3cfba 0%, #e8f5e9 100%);
|
184 |
+
border-radius: 15px;
|
185 |
+
}
|
186 |
+
.gradio-inputs, .gradio-outputs {
|
187 |
+
margin: 20px;
|
188 |
+
padding: 20px;
|
189 |
+
background-color: #ffffff;
|
190 |
+
border-radius: 10px;
|
191 |
+
box-shadow: 0px 4px 8px rgba(0, 0, 0, 0.1);
|
192 |
+
}
|
193 |
+
"""
|
194 |
)
|
195 |
)
|
196 |
|
197 |
if __name__ == "__main__":
|
198 |
+
iface.launch()
|