https://github.com/betaveros/riffshuffle

a flexible interface for interactive automatic harmonization
https://github.com/betaveros/riffshuffle

Last synced: about 6 hours ago
JSON representation

a flexible interface for interactive automatic harmonization

• Host: GitHub
• URL: https://github.com/betaveros/riffshuffle
• Owner: betaveros
• Created: 2020-05-12T06:50:04.000Z (over 4 years ago)
• Default Branch: master
• Last Pushed: 2023-03-05T13:02:04.000Z (over 1 year ago)
• Last Synced: 2023-04-10T03:22:27.443Z (over 1 year ago)
• Language: Python
• Size: 3.92 MB
• Stars: 7
• Watchers: 2
• Forks: 1
• Open Issues: 17
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
RiffShuffle

===========

My 2020 MEng thesis project, advised by Eran Egozy

The thesis is accessible on DSpace@MIT here: https://dspace.mit.edu/handle/1721.1/127386

Don't expect the code to be too good, it was all written under time pressure

## Setup

Versions of things are very not extensively tested.

### setup and data preprocessing (run this once)

- set up and activate a virtualenv with Python 3, probably Python 3.6 at least?

- `pip install -r requirements.txt`

- run `fetch.sh` to hopefully get all the data from the right places; if something fails, read this script and see if you can track the data down???

- run `parse_all.py` to parse everything into our format and store it in pickle. ~~Nottingham is very slow, like a few minutes; the others should be a few seconds each.~~ Everything is fast now because I upgraded my dependencies? ¯\\_(ツ)\_/¯

### server/client

- Server: with the virtualenv active, `python server.py`

- Client: `npm install; npm start` (`yarn` will probably work too (I forgot which dependency manager I've been using in which project, I guess this one was `npm`))

(The computations are simple enough that they could probably be done directly on the client in a WebWorker or something. I did a server/client architecture originally because I wanted to leave the door open to use more advanced machine learning libraries on the backend. That didn't happen, but it's too late now. I mean, I could probably sit down for a few hours to a few days and port all the logic to JavaScript if I felt like it, but...)

### data

See `usertests`. We lost several harmonizations because users selected the wrong melody or uploaded the wrong file: 1-2-3, 2-1-5, 2-4-1 (test number - song number - participant number). Should probably have made the server upload them itself, oops.

We also reordered the songs when presenting them to users in test 2 to get more varied data. The JSON files provided here are in the order used in test 1 and described in the thesis; in the second test, they were presented in the order song1, song4, song2, song3.