Create app.py

app.py

@@ -0,0 +1,57 @@
+import gradio as gr
+from transformers import pipeline
+import soundfile as sf
+from huggingface_hub import InferenceClient
+
+# Initialize Facebook MMS ASR model
+asr_model = pipeline("automatic-speech-recognition", model="facebook/mms-1b-all")
+
+# Initialize Facebook MMS TTS model
+tts_model = pipeline("text-to-speech", model="facebook/mms-tts")
+
+# Initialize the Chat Model (Gemma-2-9B or Futuresony.gguf)
+chat_client = InferenceClient("Futuresony/future_ai_12_10_2024.gguf")  # Change if needed
+
+def asr_chat_tts(audio):
+    """
+    1. Convert Speech to Text (ASR)
+    2. Process text through Chat Model (LLM)
+    3. Convert response to Speech (TTS)
+    """
+    # Step 1: Transcribe speech using Facebook MMS ASR
+    transcription = asr_model(audio)["text"]
+
+    # Step 2: Process text through the chat model
+    messages = [{"role": "system", "content": "You are a helpful AI assistant."}]
+    messages.append({"role": "user", "content": transcription})
+
+    response = ""
+    for msg in chat_client.chat_completion(messages, max_tokens=512, stream=True):
+        token = msg.choices[0].delta.content
+        response += token
+
+    # Step 3: Convert response to speech using Facebook MMS TTS
+    speech = tts_model(response)
+    output_file = "generated_speech.wav"
+    sf.write(output_file, speech["audio"], samplerate=speech["sampling_rate"])
+
+    return transcription, response, output_file
+
+# Gradio Interface
+with gr.Blocks() as demo:
+    gr.Markdown("<h2 style='text-align: center;'>ASR → Chatbot → TTS</h2>")
+    
+    with gr.Row():
+        audio_input = gr.Audio(source="microphone", type="filepath", label="🎤 Speak Here")
+        text_transcription = gr.Textbox(label="📝 Transcription", interactive=False)
+        text_response = gr.Textbox(label="🤖 Chatbot Response", interactive=False)
+        audio_output = gr.Audio(label="🔊 Generated Speech")
+
+    submit_button = gr.Button("Process Speech 🔄")
+
+    submit_button.click(fn=asr_chat_tts, inputs=[audio_input], outputs=[text_transcription, text_response, audio_output])
+
+# Run the App
+if __name__ == "__main__":
+    demo.launch()
+