Skip to main content

A flexible toolkit to enable replicable subjective evaluations

Project description

Replikant

**The toolkit is useable, but still requires a bit of cleaning.** This is will be the priority after Interspeech, SSW & Blizzard.

Replikant is a python toolkit which aims to enable researcher to develop their own subjective evaluation recipes. It relies on Flask, Jinja2 and SQLite, and is a refactoring of FlexEval.

I am currently doing some final important cleaning part which I thought would be done by now, but requires a bit more time.

How to install

Replikant has been tested on python 3.11.

To install simply run the following command:

pip install git+https://github.com/seblemaguer/replikant.git

When the toolkit will be fully cleaned, it will be submitted to pypi.

Developping and running a recipe

Defining an recipe

The easiest is to start with one of the available recipes and adapt it to your needs.

Here is a list of already available recipes of published studies:

Launching a recipe

Simply run :

replikant <path_configuration_recipe.yaml>

By default, the evaluation will be available at: http://127.0.0.1:8080.

The overall behaviour of replikant can be controlled from the command call. Here are the options:

usage: replikant [-h] [-d] [-i IP] [-p PORT] [-P] [-t] [-u URL] [-l LOG_FILE] [-v] RECIPE_CONFIGURATION

Replikant

positional arguments:
  RECIPE_CONFIGURATION  Recipe's configuration file

options:
  -h, --help            show this help message and exit
  -d, --debug           Start the server in debugging mode
  -i IP, --ip IP        IP's server
  -p PORT, --port PORT  port
  -P, --production      Start the server in production mode
  -t, --threaded        Enable threads.
  -u URL, --url URL     URL of the server (needed for flask redirections!) if different from http://<ip>:<port>/
  -l LOG_FILE, --log_file LOG_FILE
                        Logger file
  -v, --verbosity       increase output verbosity

Contributing

If you want to participate to the development, you can install the necessary packages using:

pip install -e .[dev]

You then need to activate the pre-commit git hooks:

pre-commit install

Citing

@inproceedings{lemaguer25_interspeech,
  title     = {Enabling the replicability of speech synthesis perceptual evaluations},
  author    = {Sébastien {Le Maguer} and Gwénolé Lecorvé and Damien Lolive and Naomi Harte and Juraj Šimko},
  year      = {2025},
  booktitle = {Interspeech 2025},
  pages     = {2545--2549},
  doi       = {10.21437/Interspeech.2025-401},
  issn      = {2958-1796},
}

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

replikant-0.0.1.tar.gz (67.2 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

replikant-0.0.1-py3-none-any.whl (90.6 kB view details)

Uploaded Python 3

File details

Details for the file replikant-0.0.1.tar.gz.

File metadata

  • Download URL: replikant-0.0.1.tar.gz
  • Upload date:
  • Size: 67.2 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.11.13

File hashes

Hashes for replikant-0.0.1.tar.gz
Algorithm Hash digest
SHA256 35d7a250ff7e024c40f7ec0428edc511fc96c6f0fd1acdcce95dc3c12b237d32
MD5 ae237adf031611552126009dab8a930b
BLAKE2b-256 4d9d03e750c8ee38772c5273e18ff928041e7021f7a1e16d9d96a5f9b0aff88b

See more details on using hashes here.

File details

Details for the file replikant-0.0.1-py3-none-any.whl.

File metadata

  • Download URL: replikant-0.0.1-py3-none-any.whl
  • Upload date:
  • Size: 90.6 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.11.13

File hashes

Hashes for replikant-0.0.1-py3-none-any.whl
Algorithm Hash digest
SHA256 0da46a237ea31425e8015738b888bf2fcfb041126df224c1ceac6acafc58fbb9
MD5 19250f842d599637f5a91172176fe65a
BLAKE2b-256 e8471aaa944160700d3c99b1c63c45eefcc0a8725542eb49cfcd554eb62c2e0f

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page