Upload 4 files

app/agent.py

@@ -0,0 +1,11 @@
+from models.local_llm import run_llm
+
+conversation_memory = []
+
+def process_text(input_text: str) -> str:
+    conversation_memory.append({"user": input_text})
+    context = "\n".join([f"User: {m['user']}" for m in conversation_memory])
+    prompt = f"You are a telecom AI assistant. Context:\n{context}\nRespond:"
+    response = run_llm(prompt)
+    conversation_memory.append({"assistant": response})
+    return response

app/main.py

@@ -0,0 +1,34 @@
+from fastapi import FastAPI, UploadFile, File, Request
+from fastapi.middleware.cors import CORSMiddleware
+from app.agent import process_text
+from app.speech_to_text import transcribe_audio
+from app.text_to_speech import synthesize_speech
+from fastapi.responses import StreamingResponse, JSONResponse
+import io
+
+app = FastAPI()
+
+app.add_middleware(
+    CORSMiddleware,
+    allow_origins=["*"],
+    allow_methods=["*"],
+    allow_headers=["*"],
+)
+
+@app.post("/transcribe")
+async def transcribe(file: UploadFile = File(...)):
+    audio_bytes = await file.read()
+    text = transcribe_audio(audio_bytes)
+    return {"transcription": text}
+
+@app.post("/query")
+async def query_agent(request: Request):
+    data = await request.json()
+    input_text = data.get("input_text", "")
+    response = process_text(input_text)
+    return {"response": response}
+
+@app.get("/speak")
+async def speak(text: str):
+    audio = synthesize_speech(text)
+    return StreamingResponse(io.BytesIO(audio), media_type="audio/wav")

app/speech_to_text.py

@@ -0,0 +1,9 @@
+import whisper
+
+model = whisper.load_model("base")
+
+def transcribe_audio(audio_bytes):
+    with open("temp.wav", "wb") as f:
+        f.write(audio_bytes)
+    result = model.transcribe("temp.wav")
+    return result["text"]

app/text_to_speech.py

@@ -0,0 +1,11 @@
+import asyncio
+import edge_tts
+
+async def generate_tts(text: str):
+    communicate = edge_tts.Communicate(text, "en-US-JennyNeural")
+    await communicate.save("speech.mp3")
+    with open("speech.mp3", "rb") as f:
+        return f.read()
+
+def synthesize_speech(text):
+    return asyncio.run(generate_tts(text))