I have built a deep learning model with python using keras (LSTM classifier for emotion speech recognition) and it’s works perfectly locally, but when i load the model or pass an audio to classify i got a high memory usage (no problem so far, because i have 16GB in my machine). I trying to deploy my model to Azure inside a flask web application and my question is: I will need a “super machine” in cloud to for run my model or there is a other method to deploy it reducing the costs?
Advertisement
Answer
You can consider leveraging Azure Machine Learning service (PaaS). It has a pay-as-you-go pricing model (per hour). Refer this for pricing based on your needed machine size. Also refer documentation.