diff --git a/readme.md b/readme.md
index c06aac559fe154ed059156e7a807a70b863226a3..3c0b6c3ad460204aa4801a3cd0e399155fa46296 100644
--- a/readme.md
+++ b/readme.md
@@ -2,6 +2,8 @@
 
 ForestMEvBat is the repository containing the code base produced for the Thomas ANTOINE's master thesis. The project is based on [ForestMEv2-public](https://forge.uclouvain.be/nbrusselmans/forestmev2-public) by Nicolas BRUSSELMANS. 
 
+All of the data needed to train the models is available at https://doi.org/10.14428/DVN/92ZO97.
+
 The project is separated between 4 folders:
 - [Embedded_sw](Embedded_sw): the STM project to program the board
 - [Annotation_generation](Annotation_generation): transform raw binary data into a usable sound clip dataset and generate annotations
@@ -38,4 +40,4 @@ A typical workflow to use the framework would be:
 
 Most scripts have multiple parameters which can be used to customize execution. Run ```script.py --help``` to see a list with a description.
 
-Note that wavs that compose the dataset will be store in a single "wavs" folder. In our examples, we will use "G:/" as path but any valid path is accepted. In our example, there should thus be a valid "G:/wavs".
\ No newline at end of file
+Note that wavs that compose the dataset will be store in a single "wavs" folder. In our examples, we will use "G:/" as path but any valid path is accepted. In our example, there should thus be a valid "G:/wavs".