Docker

docker/Dockerfile

@@ -0,0 +1,61 @@
+FROM nvcr.io/nvidia/pytorch:19.12-py3
+
+WORKDIR "/workspace"
+
+ARG DEBIAN_FRONTEND=noninteractive
+
+RUN apt-get update -y 
+RUN apt-get install -y vim git  ffmpeg libportaudio2 cmake python3.6-dev python3.7 python3.7-dev \
+	ffmpeg libsndfile1 lame cmake openssh-server python3-pyqt5 xauth python3-pip \
+	pulseaudio osspd-pulseaudio pulseaudio-module-jack gstreamer1.0-pulseaudio \
+	pulseaudio-dlna pulseaudio-esound-compat libgstreamer-plugins-base1.0-0 \
+	libgstreamer-plugins-good1.0-0 libgstreamer-plugins-bad1.0-0 liballegro-acodec5.2 libavcodec57 \
+	python-translitcodec xmms2-plugin-avcodec xubuntu-restricted-extras python-soundfile \
+	libsndifsdl2-dev python3-soundfile python-soundfile-doc \
+        && apt-get -y autoremove
+
+
+RUN python -m pip install matplotlib # ==2.1.0
+RUN python -m pip install inflect # ==0.2.5
+RUN python -m pip install librosa # ==0.6.0
+RUN python -m pip install scipy # ==1.0.0
+RUN python -m pip install tensorboardX # ==1.1
+RUN python -m pip install Unidecode # ==1.0.22
+RUN python -m pip install pillow
+RUN python -m pip install nltk # ==3.4.5
+RUN python -m pip install jamo # ==0.4.1
+RUN python -m pip install music21
+
+RUN conda install tensorflow\<2.0.0
+
+RUN apt-get install -y musescore
+
+RUN rm -Rf /root/.cache/*
+
+RUN apt-get clean
+
+RUN systemd-tmpfiles --create
+
+RUN mkdir /run/sshd
+
+RUN mkdir ./logs
+RUN chmod 777 ./logs
+
+RUN touch ./add_user.sh
+RUN echo 'nohup tensorboard --logdir /workspace/logs --host 0.0.0.0 &' >> add_user.sh
+RUN echo 'adduser --quiet --disabled-password --gecos "$1,,," --uid $2 $1' >> add_user.sh
+RUN echo 'echo "export PATH=/usr/local/nvm/versions/node/v13.3.0/bin:\$PATH" >> /home/${1}/.bashrc' >> add_user.sh
+RUN echo 'echo "export PATH=/opt/conda/bin:/opt/cmake-3.14.6-Linux-x86_64/bin/:\$PATH" >> /home/${1}/.bashrc' >> add_user.sh
+RUN echo 'echo "export PATH=/usr/local/mpi/bin:/usr/local/nvidia/bin:/usr/local/cuda/bin:\$PATH" >> /home/${1}/.bashrc' >> add_user.sh
+RUN echo 'echo Added user ${1} ${2}' >> add_user.sh
+RUN echo 'login -f $1' >> add_user.sh
+RUN chmod +x add_user.sh
+RUN sed -i 's/"\$\@"/\$\@/' /usr/local/bin/nvidia_entrypoint.sh
+
+EXPOSE 8888/tcp
+EXPOSE 6006/tcp
+
+CMD ["bash -c"]
+
+
+

docker/README.md

@@ -0,0 +1,18 @@
+# This is a Dockerfile and runscript to help you get going quickly
+- It makes container run as you instad of root
+- Automatically starts Tensorboard
+## It makes the following assumptions:
+* You are running Ubuntu
+* You have a projects folder under your Home folder
+* Under your projects folder you have cloned Mellotron
+* Also under your projects folder you have un-tarred the LibriTTS and or LJSpeech
+## Running the script:
+To run the container provide the tag you gave your image when you built it.
+...
+pneumoman@zombunnie:~/projects/mellotron$  ./run_mellotron.sh mellotronix
+...
+## Running Training
+Use full paths! I.E.
+...
+pneumoman@2a57e3b56769:/workspace/mellotron$ python train.py --output_directory=/workspace/mellotron/models --log_directory=/workspace/logs
+...

docker/run_mellotron.sh

@@ -0,0 +1,15 @@
+#!/bin/bash
+
+if [ 1 -eq $# ]
+then
+   imageid=$1
+   export GID=$(id -g)
+   sudo docker container run --rm -it --gpus all \
+       	-v ${HOME}/projects/mellotron:/workspace/mellotron \
+      	-v ${HOME}/projects/LibriTTS/train-clean-100:/path_to_libritts \
+      	-v ${HOME}/projects/LJSpeech-1.1/wavs:/path_to_ljs \
+      	-p 9876:8888 -p 9875:6006 \
+	      ${imageid} /workspace/add_user.sh ${USER} ${UID}
+else
+  echo "usage: ${0} <image name>"
+fi