Create app.py

app.py

@@ -0,0 +1,127 @@
+import subprocess
+import os
+import threading
+import numpy as np
+import librosa
+import gradio as gr
+from functools import lru_cache
+from transformers import pipeline
+from diffusers import StableDiffusionPipeline, DPMSolverMultistepScheduler
+import torch
+from huggingface_hub import login
+
+# Install required dependencies
+def install_missing_packages():
+    required_packages = {
+        "librosa": None,
+        "diffusers": ">=0.14.0",
+        "gradio": ">=3.35.2",
+        "huggingface_hub": None,
+        "accelerate": ">=0.20.1",
+        "transformers": ">=4.31.0",
+        "torch": ">=1.11.0"
+    }
+    for package, version in required_packages.items():
+        try:
+            __import__(package)
+        except ImportError:
+            package_name = f"{package}{version}" if version else package
+            subprocess.check_call(["pip", "install", package_name])
+
+install_missing_packages()
+
+# Hugging Face token authentication
+hf_token = os.getenv("HF_TOKEN")
+if hf_token:
+    login(hf_token)
+else:
+    raise ValueError("HF_TOKEN environment variable not set.")
+
+# Load the speech-to-text model
+speech_to_text = pipeline("automatic-speech-recognition", model="openai/whisper-tiny")
+
+# Load Stable Diffusion model
+text_to_image = StableDiffusionPipeline.from_pretrained(
+    "runwayml/stable-diffusion-v1-5", 
+    torch_dtype=torch.float16
+)
+device = "cuda" if torch.cuda.is_available() else "cpu"
+text_to_image.to(device)
+text_to_image.enable_attention_slicing()  # Optimizes memory usage
+text_to_image.safety_checker = None  # Disables safety checker
+text_to_image.scheduler = DPMSolverMultistepScheduler.from_config(text_to_image.scheduler.config)
+
+# Preprocess audio file into NumPy array
+def preprocess_audio(audio_path):
+    try:
+        audio, sr = librosa.load(audio_path, sr=16000)  # Resample to 16kHz
+        return np.array(audio, dtype=np.float32)
+    except Exception as e:
+        return f"Error in preprocessing audio: {str(e)}"
+
+# Transcribe audio to text
+@lru_cache(maxsize=10)
+def transcribe_audio(audio_path):
+    try:
+        audio_array = preprocess_audio(audio_path)
+        if isinstance(audio_array, str):  # Error message from preprocessing
+            return audio_array
+        result = speech_to_text(audio_array)
+        return result["text"]
+    except Exception as e:
+        return f"Error in transcription: {str(e)}"
+
+# Generate image from text
+@lru_cache(maxsize=10)
+def generate_image_from_text(text):
+    try:
+        image = text_to_image(text, height=512, width=512).images[0]
+        return image
+    except Exception as e:
+        return f"Error in image generation: {str(e)}"
+
+# Process audio input (speech-to-image)
+def speech_to_image(audio_path):
+    transcription = transcribe_audio(audio_path)
+    if "Error" in transcription:
+        return None, f"Transcription failed: {transcription}"
+    
+    image = generate_image_from_text(transcription)
+    if isinstance(image, str) and "Error" in image:
+        return None, f"Image generation failed: {image}"
+    
+    return image
+
+# Process text input (text-to-image)
+def text_to_image_interface(input_text):
+    try:
+        image = generate_image_from_text(input_text)
+        return image
+    except Exception as e:
+        return f"Error: {str(e)}"
+
+# Gradio interface
+speech_to_image_interface = gr.Interface(
+    fn=speech_to_image,
+    inputs=gr.Audio(type="filepath", label="Upload audio file (WAV/MP3)"),
+    outputs=gr.Image(label="Generated Image"),
+    title="Speech-to-Image Generator",
+    description="Upload an audio file to generate an image based on the transcribed speech."
+)
+
+text_to_image_interface = gr.Interface(
+    fn=text_to_image_interface,
+    inputs=gr.Textbox(label="Enter Text", placeholder="Describe an image..."),
+    outputs=gr.Image(label="Generated Image"),
+    title="Text-to-Image Generator",
+    description="Enter text to generate an image based on the description."
+)
+
+# Combine interfaces into a single Gradio app
+app = gr.TabbedInterface(
+    interface_list=[speech_to_image_interface, text_to_image_interface],
+    tab_names=["Speech-to-Image", "Text-to-Image"]
+)
+
+# Launch Gradio interface
+app.launch(debug=True, share=True)