Docker images for Essentia

We provide docker images for the latest version of Essentia, which contain examples (command-line feature extractors), python bindings, vamp plugins and tensorflow-support.

https://hub.docker.com/r/mtgupf/essentia/

https://hub.docker.com/r/mtgupf/essentia-nightly/

https://hub.docker.com/r/mtgupf/essentia-tensorflow/

Versions

Latest Essentia release

Images using the following base images are available:

ubuntu:20.04 (latest ubuntu LTS)
ubuntu:21.04 (latest ubuntu)
debian:bullseye (debian "stable")

The current available tags are:

ubuntu21.04-v2.1_beta5, latest (Dockerfile.ubuntu21.04-v2.1_beta5)
ubuntu20.04-v2.1_beta5 (Dockerfile.ubuntu20.04-v2.1_beta5)
bullseye-v2.1_beta5 (Dockerfile.bullseye-v2.1_beta5)

These images are based off the v2.1_beta5 release of the main essentia source tree.

Git Nightly

For a more up-to-date version of essentia we provide the mtgupf/essentia-nightly image. This is updated periodically based on the most recent commit on the master branch of essentia. We provide the latest tag, based on Ubuntu 20.04 LTS:

latest (Dockerfile.ubuntu20.04-nightly)

Tensorflow support

The latest essentia have tensorflow-support in order to easily integrate Machine Learning models into the audio processing and analysis pipelines. More details about this feature can be found here and here. We provide mtgupf/essentia-tensorflow image which contains python3 bindings and examples(command-line feature extractors with tensorflow support). This will be based on the most recent commit on the master branch of essentia. We provide the latest tag based on Ubuntu 18.04 LTS with Tensorflow 1.15.0

latest (Dockerfile.ubuntu18.04-tensorflow-v1.15.0)

Python 2

We no longer officially support Python 2, however we include a Dockerfile to build Essentia with Python 2 on Ubuntu 18.04 LTS in the python2 directory.

Usage examples

Music extractor

Analyze a file audio.wav located in the current directory and write results to the same directory:

docker run -ti --rm -v `pwd`:/essentia mtgupf/essentia essentia_streaming_extractor_music audio.wav audio.sig

Essentia in python

Run a python script using Essentia located in the current directory:

docker run -ti --rm -v `pwd`:/essentia mtgupf/essentia python3 test.py

Image management

You can rebuild an image and tag it locally using Make:

make <tag>
make ubuntu20.04-v2.1_beta5
make build # to build all images
make push # to push all images

Atemndobs/essentia