Update app.py

app.py

@@ -32,9 +32,31 @@ def text2speech(model, text, voice):
         speech = synthesiser(text, forward_params={"speaker_embeddings": speaker_embedding})
         audio_data = np.frombuffer(speech["audio"], dtype=np.float32)
         audio_data_16bit = (audio_data * 32767).astype(np.int16)
-    
         return speech["sampling_rate"], audio_data_16bit
 
+def ImageGenFromText(text, model):
+    api_key = os.getenv("fluxauthtoken")
+    login(token=api_key)
+    
+    if len(text) > 0:
+        dtype = torch.bfloat16
+        device = "cuda" if torch.cuda.is_available() else "cpu"
+        MAX_SEED = np.iinfo(np.int32).max
+        seed = random.randint(0, MAX_SEED)
+        pipe = DiffusionPipeline.from_pretrained(model, torch_dtype=dtype).to(device)
+        generator = torch.Generator().manual_seed(seed)
+        image = pipe(
+                prompt = text, 
+                width = 512,
+                height = 512,
+                num_inference_steps = 4, 
+                generator = generator,
+                guidance_scale=0.0
+        ).images[0]
+        print(image)
+        return image
+
+
 radio1 = gr.Radio(["microsoft/resnet-50", "google/vit-base-patch16-224", "apple/mobilevit-small"], value="microsoft/resnet-50", label="Select a Classifier", info="Image Classifier")
 tab1 = gr.Interface(
     fn=guessanImage,
@@ -57,5 +79,12 @@ tab3 = gr.Interface(
     outputs=["audio"],
 )
 
-demo = gr.TabbedInterface([tab1, tab2, tab3], ["tab1", "tab2", "tab3"])
+radio3 = gr.Radio(["black-forest-labs/FLUX.1-schnell"], value="black-forest-labs/FLUX.1-schnell", label="Select", info="text to image")
+tab4 = gr.Interface(
+    fn=ImageGenFromText,
+    inputs=["text", "model"],
+    outputs=["image"],
+)
+
+demo = gr.TabbedInterface([tab1, tab2, tab3, tab4], ["tab1", "tab2", "tab3", "tab4"])
 demo.launch()