Create app.py

app.py

@@ -0,0 +1,33 @@
+# app.py — Space 5
+# requirements.txt: transformers, torch, gradio, TTS, numpy, soundfile
+
+import gradio as gr
+from transformers import pipeline
+from TTS.api import TTS
+import numpy as np
+
+asr = pipeline("automatic-speech-recognition", model="E-motionAssistant/mms-300m-multilingual-ser")
+
+tts_english = TTS(model_name="E-motionAssistant/text-to-speech-VITS-english", progress_bar=False)
+tts_sinhala = TTS(model_name="E-motionAssistant/Text-to-speech-VITS-sinhala", progress_bar=False)
+tts_tamil   = TTS(model_name="E-motionAssistant/text-to-speech-VITS-tamil",   progress_bar=False)
+
+def transcribe(audio):
+    return asr(audio)["text"]
+
+def speak_english(text): return (22050, np.array(tts_english.tts(text)))
+def speak_sinhala(text): return (22050, np.array(tts_sinhala.tts(text)))
+def speak_tamil(text):   return (22050, np.array(tts_tamil.tts(text)))
+
+with gr.Blocks() as demo:
+    gr.TabbedInterface(
+        [
+            gr.Interface(fn=transcribe,    inputs=gr.Audio(type="filepath"), outputs=gr.Textbox(), title="ASR"),
+            gr.Interface(fn=speak_english, inputs=gr.Textbox(), outputs=gr.Audio(),                title="TTS English"),
+            gr.Interface(fn=speak_sinhala, inputs=gr.Textbox(), outputs=gr.Audio(),                title="TTS Sinhala"),
+            gr.Interface(fn=speak_tamil,   inputs=gr.Textbox(), outputs=gr.Audio(),                title="TTS Tamil"),
+        ],
+        ["ASR (MMS)", "TTS English", "TTS Sinhala", "TTS Tamil"]
+    )
+
+demo.launch()