File size: 10,136 Bytes
df80fb7 201ed31 103eb2f 201ed31 17454f4 201ed31 17454f4 201ed31 17454f4 201ed31 17454f4 201ed31 17454f4 201ed31 17454f4 201ed31 17454f4 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 |
import streamlit as st
import pandas as pd
import numpy as np
import os
import time
import matplotlib.pyplot as plt
from datetime import datetime
import tempfile
import io
import json
from model.transcriber import transcribe_audio
from predict import predict_emotion
# You'll need to install this package:
# pip install streamlit-audiorec
from st_audiorec import st_audiorec
AUDIO_WAV = 'audio/wav'
MAX_FILE_SIZE_MB = 10
# Page configuration
st.set_page_config(
page_title="Emotional Report Analyzer",
page_icon="🎤",
layout="wide"
)
# Initialize session state variables if they don't exist
if 'audio_data' not in st.session_state:
st.session_state.audio_data = []
if 'current_audio_index' not in st.session_state:
st.session_state.current_audio_index = -1
if 'audio_history_csv' not in st.session_state:
# Define columns for our CSV storage
st.session_state.audio_history_csv = pd.DataFrame(
columns=['timestamp', 'file_path', 'transcription', 'emotion', 'probabilities']
)
if 'needs_rerun' not in st.session_state:
st.session_state.needs_rerun = False
# Function to ensure we keep only the last 10 entries
def update_audio_history(new_entry):
# Add the new entry
st.session_state.audio_history_csv = pd.concat([st.session_state.audio_history_csv, pd.DataFrame([new_entry])], ignore_index=True)
# Keep only the last 10 entries
if len(st.session_state.audio_history_csv) > 10:
st.session_state.audio_history_csv = st.session_state.audio_history_csv.iloc[-10:]
# Save to CSV
st.session_state.audio_history_csv.to_csv('audio_history.csv', index=False)
# Function to process audio and get results
def process_audio(audio_path):
try:
# Get transcription
transcription = transcribe_audio(audio_path)
# Get emotion prediction
predicted_emotion, probabilities = predict_emotion(audio_path)
# Update audio history
new_entry = {
'timestamp': datetime.now().strftime("%Y-%m-%d %H:%M:%S"),
'file_path': audio_path,
'transcription': transcription,
'emotion': predicted_emotion,
'probabilities': str(probabilities) # Convert dict to string for storage
}
update_audio_history(new_entry)
# Update current index
st.session_state.current_audio_index = len(st.session_state.audio_history_csv) - 1
return transcription, predicted_emotion, probabilities
except Exception as e:
st.error(f"Error processing audio: {str(e)}")
return None, None, None
# Function to split audio into 10-second segments
def split_audio(audio_file, segment_length=10):
# This is a placeholder - in a real implementation, you'd use a library like pydub
# to split the audio file into segments
st.warning("Audio splitting functionality is a placeholder. Implement with pydub or similar library.")
# For now, we'll just return the whole file as a single segment
return [audio_file]
# Function to display emotion visualization
def display_emotion_chart(probabilities):
emotions = list(probabilities.keys())
values = list(probabilities.values())
fig, ax = plt.subplots(figsize=(10, 5))
bars = ax.bar(emotions, values, color=['red', 'gray', 'green'])
# Add data labels on top of bars
for bar in bars:
height = bar.get_height()
ax.text(bar.get_x() + bar.get_width()/2., height + 0.02,
f'{height:.2f}', ha='center', va='bottom')
ax.set_ylim(0, 1.1)
ax.set_ylabel('Probability')
ax.set_title('Emotion Prediction Results')
st.pyplot(fig)
# Trigger rerun if needed (replaces experimental_rerun)
if st.session_state.needs_rerun:
st.session_state.needs_rerun = False
st.rerun() # Using st.rerun() instead of experimental_rerun
col_logo, col_name = st.columns([3, 1])
col_logo.image("./img/logo_01.png", width=400)
col_name.title("Emotional Report")
# Create two columns for the main layout
col1, col2 = st.columns([1, 1])
with col1:
st.header("Audio Input")
# Method selection
tab1, tab2 = st.tabs(["Record Audio", "Upload Audio"])
with tab1:
st.write("Record your audio (max 10 seconds):")
# Using streamlit-audiorec for better recording functionality
wav_audio_data = st_audiorec()
if wav_audio_data is not None:
# Save the recorded audio to a temporary file
with tempfile.NamedTemporaryFile(delete=False, suffix='.wav') as tmp_file:
tmp_file.write(wav_audio_data)
tmp_file_path = tmp_file.name
st.success("Audio recorded successfully!")
# Process button
if st.button("Process Recorded Audio"):
# Process the audio
with st.spinner("Processing audio..."):
transcription, emotion, probs = process_audio(tmp_file_path)
# Set flag for rerun instead of calling experimental_rerun
if transcription is not None:
st.success("Audio processed successfully!")
st.session_state.needs_rerun = True
with tab2:
uploaded_file = st.file_uploader("Upload an audio file (WAV format)", type=['wav'])
if uploaded_file is not None and uploaded_file.type == AUDIO_WAV and uploaded_file.size < MAX_FILE_SIZE_MB * 1_000_000:
try:
# Save the uploaded file to a temporary location
with tempfile.NamedTemporaryFile(delete=False, suffix='.wav') as tmp_file:
tmp_file.write(uploaded_file.getbuffer())
tmp_file_path = tmp_file.name
except Exception as e:
st.error(f"Error saving uploaded file: {str(e)}")
st.error(f"Try to record your voice directly, maybe your storage is locked.")
st.audio(uploaded_file, format="audio/wav")
# Process button
if st.button("Process Uploaded Audio"):
# Split audio into 10-second segments
with st.spinner("Processing audio..."):
segments = split_audio(tmp_file_path)
# Process each segment
for i, segment_path in enumerate(segments):
st.write(f"Processing segment {i+1}...")
transcription, emotion, probs = process_audio(segment_path)
# Set flag for rerun instead of calling experimental_rerun
st.success("Audio processed successfully!")
st.session_state.needs_rerun = True
# Audio History and Analytics Section
st.header("Audio History and Analytics")
if len(st.session_state.audio_history_csv) > 0:
# Display a select box to choose from audio history
timestamps = st.session_state.audio_history_csv['timestamp'].tolist()
selected_timestamp = st.selectbox(
"Select audio from history:",
options=timestamps,
index=len(timestamps) - 1 # Default to most recent
)
# Update current index when selection changes
selected_index = st.session_state.audio_history_csv[
st.session_state.audio_history_csv['timestamp'] == selected_timestamp
].index[0]
# Only update if different
if st.session_state.current_audio_index != selected_index:
st.session_state.current_audio_index = selected_index
st.session_state.needs_rerun = True
# Analytics button
if st.button("Run Analytics on Selected Audio"):
st.subheader("Analytics Results")
# Get the selected audio data
selected_data = st.session_state.audio_history_csv.iloc[selected_index]
# Display analytics (this is where you would add more sophisticated analytics)
st.write(f"Selected Audio: {selected_data['timestamp']}")
st.write(f"Emotion: {selected_data['emotion']}")
st.write(f"File Path: {selected_data['file_path']}")
# Add any additional analytics you want here
# Try to play the selected audio
try:
if os.path.exists(selected_data['file_path']):
st.audio(selected_data['file_path'], format="audio/wav")
else:
st.warning("Audio file not found - it may have been deleted or moved.")
except Exception as e:
st.error(f"Error playing audio: {str(e)}")
else:
st.info("No audio history available. Record or upload audio to create history.")
with col2:
st.header("Results")
# Display results if available
if st.session_state.current_audio_index >= 0 and len(st.session_state.audio_history_csv) > 0:
current_data = st.session_state.audio_history_csv.iloc[st.session_state.current_audio_index]
# Transcription
st.subheader("Transcription")
st.text_area("", value=current_data['transcription'], height=100, key="transcription_area")
# Emotion
st.subheader("Detected Emotion")
st.info(f"🎭 Predicted emotion: **{current_data['emotion']}**")
# Convert string representation of dict back to actual dict
try:
import ast
probs = ast.literal_eval(current_data['probabilities'])
display_emotion_chart(probs)
except Exception as e:
st.error(f"Error parsing probabilities: {str(e)}")
st.write(f"Raw probabilities: {current_data['probabilities']}")
else:
st.info("Record or upload audio to see results")
# Footer
st.markdown("---")
st.caption("Emotional Report Analyzer - Processes audio in 10-second segments and predicts emotions") |