Skip to content

AI-Unicamp/TTS-Subjective-Evaluation

 
 

Repository files navigation

webMUSHRA

GitHub release Github All Releases Build Status DOI

a MUltiple Stimuli with Hidden Reference and Anchor (MUSHRA) compliant web audio API based experiment software.

screen shot 2017-03-08 at 12 38 20

Introduction

Listening tests are widely used to assess the quality of audio systems. In the last few years, conducting listening experiments over the Internet, as so called web-based experiments, has become popular. Until now, it was only possible to implement a limited number of listening test types as web-based experiments because web standards were missing some crucial features, e.g. sample manipulation of audio streams. MUSHRA tests are designed to compare the audio quality of several test conditions with intermediate impairments to a high quality reference. With the rise of Web Audio API, for the first time MUSHRA experiments can be carried out within the web browser while at the same time being compliant to the ITU-R Recommendation BS.1534 (MUSHRA).

Download

We provide two version of webMUSHRA.

  • webMUSHRA provides the version targeted for normale usage and experimenters. The javascript files are compressed which makes it faster to load/serve. The documentation is provided as PDFs.

  • webMUSHRA-dev is targeted to developers and experienced users who want to customize experiments. This version is comparable to cloning the git repository

Download Package Here

Features

  • page based experiments supporting:
    • MUSHRA (ITU-R BS.1534)
    • AB (ITU-R BS.1116)
    • Likert scale questionaires
    • training/introduction
    • spatial attributes, such as ASW, LEV, and localization (experimental)
  • compliant to ITU recommendations (looping, fade-in/out, sample accurate switching)
  • finish page to gather the results and send them to a provided PHP service
  • client side processing using the Web Audio API
  • simple configuration using YAML preference files
  • automatically generates ITU-R compliant lower anchor files on the fly
  • keyboard shortcuts for interaction with main UI elements

Supported Browsers

  • Google Chrome on Windows, Mac and Linux

Getting started: Setting up webMUSHRA using PHP's builtin webserver

To load audio files and save the results as csv text files, webMUSHRA needs to run on a web server. If you already have php installed on your system (for example on Mac OS X), you can run a php development server on port 8000 from the terminal using php -S localhost:8000.

Now you can run webMUSHRA using the following URL: http://localhost:8000

The experiment configurations are stored in the configs/ folder. To load a configuration/experiment, specify the config argument in the url http://localhost:8000/?config=mushra_showresults.yaml.

Docker

You can use docker to set up webMUSHRA quickly. Just run docker-compose -f docker-compose.yml build to build the webMUSHRA docker container.

To run the container use webMUSHRA docker-compose -f docker-compose.yml up. We configured the docker image so that the configs and the results folder is mounted inside the container so that you can modify it on the fly and receive results within the results folder.

Note for Docker on Windows

When using Docker Toolbox/Machine on Windows, volume paths (to mount the configs and results folder) are not converted by default. To enable this conversion set the environment variable COMPOSE_CONVERT_WINDOWS_PATHS=1 e.g. by env:COMPOSE_CONVERT_WINDOWS_PATHS=1 in the power shell.

Apache + PHP

Another custom way to run webMUSHRA would be to install a complete web server stack like XAMPP.

Python Backend

A python based backend to save the results in provided by pymushra.

Change or add a configuration

webMUSHRA uses YAML to configure experiments. Since YAML is using whitespace indentation (no tab characters!), we recommend to use a text editor like Atom that ships with YAML support.

A simple MUSHRA test in YAML looks like this:

pages:
  - type: mushra
    id: Item 1
    name: Orchestra
    content: Add additional notes for the participants
    showWaveform: true
    reference: reference.wav
    createAnchor35: true
    createAnchor70: true
    stimuli:
      C1: codec1.wav
      C2: codec2.wav
      C3: codec3.wav

The specific parameters are described in the Experimenters Manual.

Documentation

Citation

If you use webMUSHRA in your publication, please cite it using the following reference:

Schoeffler, M. et al., (2018). webMUSHRA — A Comprehensive Framework for Web-based Listening Tests. Journal of Open Research Software. 6(1), p.8.

References

Copyright/Licence

(C) AudioLabs 2020

This source code is protected by copyright law and international treaties. This source code is made available to you subject to the terms and conditions of the Software License for the webMUSHRA.js Software. Said terms and conditions have been made available to you prior to your download of this source code. By downloading this source code you agree to be bound by the above mentionend terms and conditions, which can also be found here. Any unauthorised use of this source code may result in severe civil and criminal penalties, and will be prosecuted to the maximum extent possible under law.

About

a MUSHRA compliant web audio API based experiment software

Resources

License

Stars

Watchers

Forks

Packages

No packages published

Languages

  • JavaScript 84.8%
  • CSS 7.0%
  • PHP 5.3%
  • HTML 2.9%