Update app.py

app.py

@@ -3,6 +3,9 @@ import numpy as np
 import gradio as gr
 from transformers import pipeline
 import logging
+from scipy.io.wavfile import write
+import uuid
+import os
 
 # -----------------------------
 # Reduce Transformers warnings
@@ -10,7 +13,7 @@ import logging
 logging.getLogger("transformers").setLevel(logging.ERROR)
 
 # -----------------------------
-# LOAD PIPELINE (HF AUTH REQUIRED)
+# LOAD PIPELINE
 # -----------------------------
 device = 0 if torch.cuda.is_available() else -1
 tts_pipe = pipeline(
@@ -26,31 +29,25 @@ def tts_generate(text):
     if not text.strip():
         return None
 
-    try:
-        output = tts_pipe(text)
-    except Exception as e:
-        print("Error:", e)
-        return None
+    output = tts_pipe(text)
 
     audio = np.asarray(output["audio"], dtype=np.float32)
     sr = output["sampling_rate"]
 
-    return (sr, audio)
+    os.makedirs("outputs", exist_ok=True)
+    out_path = f"outputs/{uuid.uuid4().hex}.wav"
+
+    write(out_path, sr, audio)
+
+    return out_path
 
 # -----------------------------
-# SAMPLE TEXTS WITH EXPRESSIVE TAGS
+# SAMPLE TEXTS
 # -----------------------------
 SAMPLES = [
-    "Just end up crashing somewhere. <laugh> No, because remember last time? You fell asleep—",
-    "Running through the grass, playing under the falling leaves. <chuckle> My sweet little kit, the—",
-    "Deal with it. I will. I'll just scowl and watch TV by myself <sigh>.",
+    "Just end up crashing somewhere. <laugh> No, because remember last time?",
     "Hmm… I don't know. <laugh> This feels like a bad idea. <gasp>",
     "I'm so tired today <yawn> but I still have so much work to do.",
-    "<cough> Wait—did you hear that? <gasp> I swear something just moved.",
-    "<whisper> Don't turn around. Just keep walking.",
-    "Ugh… <sigh> I can't believe this is happening again.",
-    "Okay okay <laughs nervously> maybe it wasn't my best decision.",
-    "I just got a new sword, ooh, it's a mighty Anduril! <groan>"
 ]
 
 # -----------------------------
@@ -59,23 +56,14 @@ SAMPLES = [
 demo = gr.Interface(
     fn=tts_generate,
     inputs=gr.Textbox(
-        label="Enter text (use expressive tags like <laugh>, <sigh>, <gasp>)",
+        label="Enter text (use expressive tags like <laugh>, <sigh>)",
         lines=5,
         placeholder=SAMPLES[0],
     ),
-    outputs=gr.Audio(type="numpy", label="Generated Audio"),
-    title="Orpheus‑3B Expressive TTS",
-    description=(
-        "You can add expressiveness to speech by inserting tags:\n"
-        "- <laugh>, <chuckle>: For laughter sounds\n"
-        "- <sigh>: For sighing sounds\n"
-        "- <cough>, <sniffle>: For subtle interruptions\n"
-        "- <groan>, <yawn>, <gasp>: For additional emotional expression\n\n"
-        "Tags can appear at the start, middle, or end of a sentence.\n"
-        "Example: `I can't believe it! <laugh>`"
-    ),
+    outputs=gr.Audio(type="filepath", label="Generated Audio"),
+    title="Orpheus-3B Expressive TTS",
     examples=[[s] for s in SAMPLES],
 )
 
 if __name__ == "__main__":
-    demo.launch(ssr_mode=False)
+    demo.launch(enable_queue=False)