randomly sampled speech segment from EdAcc:

120 Speaker. Speakers span different English variaties (L1, and L2).
Linguistic Background. We document the linguistic profile for every speaker.
Dyadic conversations. We provide recording between two speakers, who already know each other.
Virtual Calls. We propose a new (unexplored) domain in the speech community.
No Topics Constrains. Speakers can explore any topic without vocabulary restrictions.

News

[03/23] We created a Leaderboard. We will be manually compiling results on EdAcc.
[03/23] EdAcc V1.0 released. We crearte an STM file so that evaluation can be done at the sentence level and include evaluation code for it. You can find previous version in the Leaderboard Section.
[01/23] Paper accepted to ICASSP 2023! See you in Greece :) Let's discuss future (and current) direction for EdAcc

Download

Data

Paper

Bibtex:

@inproceedings{sanabria23edacc,
   title={{T}he {E}dinburgh {I}nternational {A}ccents of {E}nglish {C}orpus: {T}owards the {D}emocratization of {E}nglish {ASR}},
   author={Sanabria, Ramon and Bogoychev, Nikolay and  Markl, Nina and Carmantini, Andrea and  Klejch, Ondrej and Bell, Peter},
   booktitle={ICASSP 2023},
   year={2023},
}

Data Statement

Contribute

You can contribute in many ways!

Do you have an interesting English accent? Record a conversation with a friend and send it to us Here is how
Interested on funding EdAcc? We are actively looking for funding, to keep transcribing data. Contact us!
Ideas for further research on EdAcc and accents? get in touch with us us well!

Papers

Here we will list publications that use EdAcc, and how they use it. We hope that this list inspire future research with EdAcc.

The Edinburgh International Accents of English Corpus: Towards the Democratization of English ASR. Ramon Sanabria, Bogoychev, Markl, Carmantini, Klejch, and Bell. ICASSP 2023. Presentation of the EdAcc.

Ethics

This project has been approved by the University of Edinburgh, Informatics Ethics Board -- Ref. 49776. It has been funded by the Institute for Language, Cognition, and Computation at the University of Edinburgh. All contributors provided informed consent to publicly share their speech recordings and demographic information. We acknowledge that despite efforts to design an accessible and fair data curation process, we are only representing a small section of all English speakers. We keep encouraging speakers from underrepresented groups to contribute to the dataset to make it more representative.