========= appension

Web player for devil glitch extension

• Infinite glitch - the longest pop song ever

Requirements

Postgresql database server (on OSX brew install postgresql) FFMPEG audio manipulation and rendering (on OSX brew install ffmpeg)

Installation

• Create a virtualenv python3 -m venv glitch

• • cd glitch
• • source bin/activate

• Install pip Requirements pip install -r requirements.txt

• Create an apikeys.py file based on apikeys_sample.py

• Generate a privkey.pem file in the main directory, above "glitch"

• • ssh-keygen -f ~/path/to/id_rsa.pub -m 'PEM' -e > privkey.pem

• Insure postgres is running (on OSX) brew services start postgresql

• • Usually the package manager creates a database for you, but if you get an error that database does not exist, try running the command createdb (reference)

• Create the database tables python -m glitch tables --confirm

• • (You can test by running this command without the --confirm flag.)

• Create a new user

• • python3 -m glitch create-user name email password

• Make new user admin

• • psql `whoami`
• • update users set status=1 where id = 1;
• • update users set user_level=2 where id = 1;

• Create an audio directory mkdir audio and put some mp3 files in it.

• Add files to glitchery python3 -m glitch importmp3 audio/*.mp3

• Visit the address chosen in apikeys for admin_address to make tracks active (or do via psql).

Start Server

python3 -m glitch main

Start Renderer

python3 -m glitch renderer

Build/Rebuild Major Glitch

python3 -m glitch major-glitch

For dev if you want to play tracks more quietly

python3 -m glitch renderer --gain=-10 (where -10 is a number of decibels)

To View All CMDLINE Commands

python -m glitch --help

On Unix system with Systemd

Use bash to run the makeservice.sh file, then start both services with command: sudo systemctl start glitch glitch-renderer

If using a service with Unicorn and using the Pike Monitor

sudo systemctl start glitch.service

check status: systemctl status glitch

run monitor like this: sudo pike outage.pike OR (chmod +x outage.pike) sudo ./outage.pike

SSL Certificates are generated via Let's Enctypt

See this link Also create a deploy hook for renewal:

/etc/letsencrypt/renewal-hooks/deploy/appension 
cp /etc/letsencrypt/live/infiniteglitch.net/fullchain.pem /etc/letsencrypt/live/infiniteglitch.net/privkey.pem /home/mikekilmer/appension

To manually renew certificate with certbot:

sudo certbot renew --webroot -w /home/mikekilmer/appension
( --dry-run --debug-challenges flags can be useful )

On a Fresh Server

In order to have services start automatically on reboot via systemd, each service must be symlinked thusly:

sudo systemctl enable glitch glitch-renderer glitch-redirect sudo systemctl start glitch glitch-renderer glitch-redirect

Environment

Running on Debian 10, Buster with Python 3.7.3

Requirements versions as of 31 July 2019

• aiohttp 3.5.4
• amen 0.0.3
• asn1crypto 0.24.0
• async-timeout 3.0.1
• attrs 19.1.0
• audioread 2.1.8
• certifi 2019.6.16
• cffi 1.12.3
• chardet 3.0.4
• Click 7.0
• clize 4.0.3
• cryptography 2.7
• DateTime 4.3
• decorator 4.4.0
• docutils 0.15.2
• Flask 1.1.1
• Flask-Login 0.4.1
• gunicorn 19.9.0
• idna 2.8
• itsdangerous 1.1.0
• Jinja2 2.10.1
• joblib 0.13.2
• librosa 0.7.0
• llvmlite 0.29.0
• MarkupSafe 1.1.1
• multidict 4.5.2
• mutagen 1.42.0
• numba 0.45.0
• numpy 1.17.0
• od 1.0
• pandas 0.25.0
• pip 18.1
• pkg-resources 0.0.0
• psycopg2-binary 2.8.3
• pycparser 2.19
• pycrypto 2.6.1
• pydub 0.23.1
• pyOpenSSL 19.0.0
• PySoundFile 0.9.0.post1
• python-dateutil 2.8.0
• pytz 2019.1
• requests 2.22.0
• resampy 0.2.1
• scikit-learn 0.21.3
• scipy 1.3.0
• setuptools 40.8.0
• sigtools 2.0.2
• simplejson 3.16.0
• six 1.12.0
• SoundFile 0.10.2
• stop-words 2018.7.23
• urllib3 1.25.3
• Werkzeug 0.15.5
• yarl 1.3.0
• zope.interface 4.6.0

Troubleshooting

• Check CPU Usage with top
• systemctl status glitch glitch-renderer
• sudo journalctl -f -u glitch
• sudo journalctl -f -u glitch-renderer

A brief History

A few years ago Chris Butler decided to extend The Devil Glitch into infinity. Dozens of artists began writing and recording verses to contribute and re-assembling the gigantic track became rather resource-intensive.

We looked for an internet Music Player with a gapless playback feature, but found none. Chris' friend Henry Lowengard suggested with do something with the Echonest API and it's python companions, PyEchoNest and Remix. I didn't know Python, but was interested in learning and we decided to forge ahead in this direction.

Found my way to Peter Sobot's Forever.fm codebase (or a version of it), forked it and started reading and tinkering. Before long I realized I was way out of my depth and dug up a mentor named Chris Angelico with whos help Infinite Glitch was born.

Echonest API has been migrated to a new API that doesn't expose the Track Analysis attributes so we need to replace it.

Minimum Viable Product Requirements

• Work with Audio Streams (as opposed to complete files)
• Insert streaming tracks --- gap-lessly --- into output stream randomly
• Write streams out to a single track

Admin

• Sequence of tracks for Single Rendered File
• Trim start time either in milliseconds or “beats” for use in transitions

This would probably give us something that works, but lacks certain current functionality.

Additional Current Features

• Crossfade between tracks
• Combine tracks --- required for the Recording Studio
• Normalize and Limit track volumes

What can Amen do?

The Amen echo_nest_converter AudioAnalysis object contains a reference to the original file as well as list of analysis data:

• sections
• bars
• beats
• tatums
• segments

It looks like sections, tatums and segments may not yet be implemented. As far as I can tell the amen.synthesize function is what renders an final audio file, but as we're dealing with streams we may need to handle differently. Possibly RealAudioStream from Forever.fm will work, or maybe some variation on a Threading object.

For combining the two elements in the Recording Studio, the PyDub might be perfect. It doesn't support streaming so not sure if we could use it to crossfade streaming PMC or any other type of audio data very easily.