[joss] Need locations of benchmarking data!

juanmc2005 / diart

A python package to build AI-powered real-time audio applications

https://diart.readthedocs.io

MIT License

1.09k stars 91 forks source link

[joss] Need locations of benchmarking data! #175

Open sneakers-the-rat opened 1 year ago

sneakers-the-rat commented 1 year ago

Trying to find the data to run the benchmarks, and I can't find all the source data:

[x] VoxConverse - found it!
[ ] AMI - found the download page https://groups.inf.ed.ac.uk/ami/download/ but I can't tell what files I need for the benchmark
[x] *DIHARD III - can't find download
[x] *DIHARD II - can't find download

this should all be in the docs!

part of https://github.com/openjournals/joss-reviews/issues/5266

sneakers-the-rat commented 1 year ago

Forgot to say - I also couldnt tell easily from the text of the repo paper which samples I should use from voxconverse, might be nice to have a link to a source and a little script for loading them - making reproducing like 1 step if you have a spare half hour or so

juanmc2005 commented 1 year ago

@sneakers-the-rat good idea! However, the two DIHARD datasets are private. One has to ask for permission to use them. For AMI we used Mix-Headset (downloadable here)