Create app.py

app.py

@@ -0,0 +1,45 @@
+import gradio as gr
+from asr import transcribe_auto  # Import ASR function
+from ttsmms import download, TTS
+from langdetect import detect
+
+# Download and load TTS models for Swahili and English
+swahili_dir = download("swh", "./data/swahili")
+english_dir = download("eng", "./data/english")
+
+swahili_tts = TTS(swahili_dir)
+english_tts = TTS(english_dir)
+
+# Function to handle ASR → TTS
+def asr_to_tts(audio):
+    # Step 1: Transcribe Speech
+    transcribed_text = transcribe_auto(audio)
+
+    # Step 2: Detect Language & Generate Speech
+    lang = detect(transcribed_text)
+    wav_path = "./output.wav"
+
+    if lang == "sw":  # Swahili
+        swahili_tts.synthesis(transcribed_text, wav_path=wav_path)
+    else:  # Default to English
+        english_tts.synthesis(transcribed_text, wav_path=wav_path)
+
+    return transcribed_text, wav_path  # Return both text & generated speech
+
+# Gradio Interface
+with gr.Blocks() as demo:
+    gr.Markdown("<h2 style='text-align: center;'>Multilingual Speech-to-Text & Text-to-Speech</h2>")
+    
+    with gr.Row():
+        audio_input = gr.Audio(source="microphone", type="filepath", label="🎤 Speak Here")
+        text_output = gr.Textbox(label="📝 Transcription", interactive=False)
+        audio_output = gr.Audio(label="🔊 Generated Speech")
+
+    submit_button = gr.Button("Transcribe & Speak 🔄")
+
+    submit_button.click(fn=asr_to_tts, inputs=[audio_input], outputs=[text_output, audio_output])
+
+# Run the App
+if __name__ == "__main__":
+    demo.launch()
+