test(tts): checking whether the tts model is working correctly

Dockerfile

@@ -1,26 +1,21 @@
-# Zamiast FROM python:3.12, użyj swojego obrazu bazowego
-FROM crazymonkey00/llama-base:latest
+# Instead of FROM python:3.12, use a slim base image
+FROM python:3.12-slim
 
-# Ustaw katalog roboczy
+# Set the working directory
 WORKDIR /app
 
-# Skopiuj requirements.txt (bez llama-cpp-python - już jest w obrazie!)
+# Copy requirements.txt (llama-cpp-python is already included in the base image)
 COPY ./requirements.txt /app/requirements.txt
 
-# Zainstaluj tylko dodatkowe zależności z Twojego projektu
+# Install only project-specific dependencies
 RUN pip install --no-cache-dir -r requirements.txt
 
-# Skopiuj cały kod aplikacji
+# Copy the entire application source code
 COPY . /app
 
-# Expose port dla Hugging Face Spaces
+# Expose port for Hugging Face Spaces
 EXPOSE 7860
 
-# set environment variables
-ENV CUDA_VISIBLE_DEVICES=""
-ENV TF_CPP_MIN_LOG_LEVEL=2
-ENV TF_ENABLE_ONEDNN_OPTS=0
-
 # Run FastAPI with Gunicorn - increased timeout for model loading
 CMD ["gunicorn", "app.main:app", \
     "-k", "uvicorn.workers.UvicornWorker", \

app/routes/nlp.py

@@ -66,23 +66,21 @@ async def chat(request: Request, chat_request: ChatRequest):
         skip_special_tokens=True
     ).strip()
 
-    return {"response": response_text}
-
-    # # Generate audio using TTS
-    # audio_bytes = save_audio(request, response_text)
-
-    # # Create a multipart response
-    # boundary = uuid.uuid4().hex
-    # body = (
-    #     f"--{boundary}\r\n"
-    #     f"Content-Disposition: form-data; name=\"text\"\r\n\r\n"
-    #     f"{response_text}\r\n"
-    #     f"--{boundary}\r\n"
-    #     f"Content-Disposition: form-data; name=\"audio\"; filename=\"speech.wav\"\r\n"
-    #     f"Content-Type: audio/wav\r\n\r\n"
-    # ).encode() + audio_bytes + f"\r\n--{boundary}--\r\n".encode()
-
-    # return Response(
-    #     content=body,
-    #     media_type=f"multipart/form-data; boundary={boundary}"
-    # )
+    # Generate audio using TTS
+    audio_bytes = save_audio(request, response_text)
+
+    # Create a multipart response
+    boundary = uuid.uuid4().hex
+    body = (
+        f"--{boundary}\r\n"
+        f"Content-Disposition: form-data; name=\"text\"\r\n\r\n"
+        f"{response_text}\r\n"
+        f"--{boundary}\r\n"
+        f"Content-Disposition: form-data; name=\"audio\"; filename=\"speech.wav\"\r\n"
+        f"Content-Type: audio/wav\r\n\r\n"
+    ).encode() + audio_bytes + f"\r\n--{boundary}--\r\n".encode()
+
+    return Response(
+        content=body,
+        media_type=f"multipart/form-data; boundary={boundary}"
+    )