Deep Audio

Setup Instructions

1. Follow the installation instructions for a tensorflow docker container: https://www.tensorflow.org/install/docker#tensorflow_docker_requirements

2. When creating the docker container run the following to create a persistant container:

   sudo docker container create \
       --name deep-audio \
       --restart always \
       -p 8888:<exposed_port> \
       -p 6006:<exposed_port> \
       -v <notebook_dir_in_home>:/notebooks \
       tensorflow/tensorflow:latest-gpu-py3

3. Once created, be sure that the port you use for jupyter notebooks is open on the server (by default 8888)

4. Start the new container through

   sudo docker start deep-audio

5. Open a bash shell into your container using

   sudo docker exec -it deep-audio /bin/bash

6. Install dependencies:

   apt update
   apt install git openssh-server openssh-client nano libsndfile-dev

7. Add ssh key to Github:

   ssh-keygen -t rsa -b 4096 -C "your_email@example.com"
   eval "$(ssh-agent -s)"
   ssh-add ~/.ssh/id_rsa
   cat ~/.ssh/id_rsa.pub

   From: https://help.github.com/articles/generating-a-new-ssh-key-and-adding-it-to-the-ssh-agent/#platform-linux

8. Clone the repository into container

   git clone --recursive git@github.com:georgia-tech-db/deep-audio.git

9. Enter the directory and install pip dependencies

   pip install -r requirements.txt

10. (Optional) Exit the container and if remote reconnect to the server with ssh tunneling:

    ssh -L 8888:localhost:8888 usr@server