What you are doing here will only be able to play the speech on server side. Instead, you will have to save the output as a audio file and then return it on client side.
There is a similar example that uses gtts instead of pyttsx3