Spaces:
Sleeping
Sleeping
File size: 9,093 Bytes
fd8dafd |
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 |
# Import necessary libraries
import torch
import numpy as np
import transformers
import scipy.io.wavfile as wavfile
import openai
from transformers import pipeline
from transformers import WhisperProcessor, WhisperForConditionalGeneration
from gtts import gTTS
import os
import gradio as gr
import librosa
# Set your OpenAI API key (consider using environment variables for security)
openai.api_key = "your_api_key_here" # Replace with your actual API key
class MoodEnhancerModel:
def __init__(self):
print("Initializing Mood Enhancer Model...")
# Initialize Whisper for speech recognition
print("Loading Whisper model...")
self.whisper_processor = WhisperProcessor.from_pretrained("openai/whisper-base")
self.whisper_model = WhisperForConditionalGeneration.from_pretrained("openai/whisper-base")
# Initialize BERT for sentiment analysis/mood detection
print("Loading BERT model...")
self.sentiment_analyzer = pipeline(
"sentiment-analysis",
model="nlptown/bert-base-multilingual-uncased-sentiment"
)
print("All models loaded successfully!")
def transcribe_audio(self, audio_file):
"""Transcribe audio using Whisper"""
print("Transcribing audio...")
# Process through Whisper API for better results
try:
with open(audio_file, "rb") as f:
audio_data = f.read()
transcript = openai.Audio.transcribe("whisper-1", audio_data)
transcribed_text = transcript["text"]
except Exception as e:
print(f"OpenAI API error: {e}")
# Fallback to local Whisper model
# Load and preprocess the audio
audio_array, sampling_rate = librosa.load(audio_file, sr=16000)
input_features = self.whisper_processor(
audio_array, sampling_rate=16000, return_tensors="pt"
).input_features
# Generate token ids
predicted_ids = self.whisper_model.generate(input_features)
# Decode token ids to text
transcribed_text = self.whisper_processor.batch_decode(
predicted_ids, skip_special_tokens=True
)[0]
print(f"Transcribed text: {transcribed_text}")
return transcribed_text
def analyze_mood(self, text):
"""Analyze mood using BERT"""
print("Analyzing mood...")
results = self.sentiment_analyzer(text)
# Convert 1-5 star rating to mood scale
sentiment_score = int(results[0]['label'].split()[0])
moods = {
1: "very negative",
2: "negative",
3: "neutral",
4: "positive",
5: "very positive"
}
detected_mood = moods[sentiment_score]
print(f"Detected mood: {detected_mood}")
return detected_mood, sentiment_score
def generate_response(self, text, mood, mood_score):
"""Generate mood enhancing response using OpenAI"""
print("Generating mood enhancing response...")
# Customize the prompt based on detected mood
if mood_score <= 2:
prompt = f"""
The user seems to be feeling {mood}. Their message was: "{text}"
Generate an empathetic and uplifting response that acknowledges their feelings
but helps shift their perspective to something more positive. Keep the response
conversational, warm and under 3 sentences.
"""
elif mood_score == 3:
prompt = f"""
The user seems to be feeling {mood}. Their message was: "{text}"
Generate a cheerful response that builds on any positive aspects of their message
and adds some encouraging thoughts. Keep the response conversational,
warm and under 3 sentences.
"""
else:
prompt = f"""
The user seems to be feeling {mood}. Their message was: "{text}"
Generate a response that celebrates their positive state and offers a way to
maintain or enhance this good feeling. Keep the response conversational,
warm and under 3 sentences.
"""
try:
# Updated for OpenAI's current API format
response = openai.ChatCompletion.create(
model="gpt-3.5-turbo",
messages=[
{"role": "system", "content": "You are an empathetic AI assistant designed to enhance the user's mood."},
{"role": "user", "content": prompt}
],
max_tokens=150,
temperature=0.7
)
enhanced_response = response['choices'][0]['message']['content'].strip()
print(f"Generated response: {enhanced_response}")
return enhanced_response
except Exception as e:
print(f"Error with OpenAI API: {e}")
# Fallback responses if API fails
if mood_score <= 2:
return "I notice you might be feeling down. Remember that challenging moments are temporary, and small positive steps can help shift your perspective."
elif mood_score == 3:
return "I sense a neutral mood. What's one small thing that brought you joy today? Focusing on positive moments, even tiny ones, can boost your overall wellbeing."
else:
return "It sounds like you're in a good mood! That's wonderful to hear. Savoring these positive feelings can help them last longer."
def text_to_speech(self, text):
"""Convert text to speech"""
print("Converting to speech...")
tts = gTTS(text=text, lang='en', slow=False)
output_path = "response.mp3"
tts.save(output_path)
return output_path
def process_text_input(self, text_input):
"""Process text input and return results"""
mood, mood_score = self.analyze_mood(text_input)
response = self.generate_response(text_input, mood, mood_score)
audio_file = self.text_to_speech(response)
return text_input, mood, mood_score, response, audio_file
def process_audio_input(self, audio_file):
"""Process audio input and return results"""
text = self.transcribe_audio(audio_file)
mood, mood_score = self.analyze_mood(text)
response = self.generate_response(text, mood, mood_score)
audio_response = self.text_to_speech(response)
return text, mood, mood_score, response, audio_response
# Initialize the model
model = MoodEnhancerModel()
# Create a Gradio interface for text input
def text_interface(text):
input_text, mood, mood_score, response, audio_file = model.process_text_input(text)
return mood, f"Mood score: {mood_score}/5", response, audio_file
# Create a Gradio interface for audio input
def audio_interface(audio):
input_text, mood, mood_score, response, audio_file = model.process_audio_input(audio)
return input_text, mood, f"Mood score: {mood_score}/5", response, audio_file
# Create Gradio tabs for different input types
with gr.Blocks(title="Mood Enhancer") as demo:
gr.Markdown("# Mood Enhancer")
gr.Markdown("Upload an audio file or enter text to analyze your mood and receive an uplifting response.")
with gr.Tabs():
with gr.TabItem("Text Input"):
with gr.Row():
text_input = gr.Textbox(label="Enter your text", placeholder="How are you feeling today?", lines=3)
text_button = gr.Button("Analyze Mood")
with gr.Row():
text_mood = gr.Textbox(label="Detected Mood")
text_score = gr.Textbox(label="Mood Score")
text_response = gr.Textbox(label="Response", lines=3)
text_audio = gr.Audio(label="Audio Response")
text_button.click(
fn=text_interface,
inputs=text_input,
outputs=[text_mood, text_score, text_response, text_audio]
)
with gr.TabItem("Audio Input"):
audio_input = gr.Audio(label="Upload or Record Audio", type="filepath")
audio_button = gr.Button("Analyze Audio")
with gr.Row():
transcribed_text = gr.Textbox(label="Transcribed Text")
with gr.Row():
audio_mood = gr.Textbox(label="Detected Mood")
audio_score = gr.Textbox(label="Mood Score")
audio_response = gr.Textbox(label="Response", lines=3)
response_audio = gr.Audio(label="Audio Response")
audio_button.click(
fn=audio_interface,
inputs=audio_input,
outputs=[transcribed_text, audio_mood, audio_score, audio_response, response_audio]
)
# Launch the Gradio interface
demo.launch(share=True) |