Skip to content


Folders and files

Last commit message
Last commit date

Latest commit



47 Commits

Repository files navigation

Multilingual Acoustic Word Embeddings on Globalphone


PyTorch implementation of multilingual acoustic word embedding approaches. The experiments are described in:

  • C. Jacobs, H. Kamper, and Y. Matusevych, "Acoustic word embeddings for zero-resource languages using self-supervised contrastive learning and multilingual adaptation," in Proc. SLT, 2021. [arXiv]

The same Contrastive RNN implementation is used in:

  • C. Jacobs and H. Kamper, "Multilingual transfer of acoustic word embeddings improves when training on languages related to the target zero-resource language," in Proc. Interspeech, 2021. [arXiv]


Use code at own risk.

Download datasets

The GlobalPhone corpus and forced alignments of the data needs to be obtained. GlobalPhone needs to be paid for. If you have proof of payment, we can give you access to the forced alignments. Save the data and forced alignments in a separate directory and update the file to point to the data directories.

Install dependencies

You will require the following:

To install speech_dtw (required for same-different evaluation) and shorten (required for processing audio), run ./

You can install all the other dependencies in a conda environment by running:

conda env create -f environment.yml
conda activate pyt1.4

Extract speech features

Update the paths in to point to the data directories. Extract MFCC features in the features/ directory as follows:

cd features
./ SP

You need to run for all languages; run it without any arguments to see all 16 language codes.

UTD pairs can also be analysed here, by running e.g.:

./ SP



PyTorch implementations of Acoustic Word Embedding models






No releases published


No packages published