TechCabal is pleased to announce the launch of the Ewè audio classification challenge in partnership with Zindi and Umbaji.
If you’re passionate about artificial intelligence and natural language processing and you want to push the boundaries of audio classification technology, we invite you to participate in this competition.
Participants in this competition will develop machine learning models capable of accurately classifying audio recordings of basic directional commands spoken in Ewè (“up,” “down,” “stop,” “go,” “left,” “right,” “yes,” and “no.”) The solutions created in this challenge will be used in a navigation voice assistant for visually impaired individuals who speak Ewè.
Navigating public spaces is a significant challenge for visually impaired individuals globally. Being able to interpret and respond to spoken directions supports the safety and independence of these individuals. With 5 million speakers in mainly Ghana, Togo, and Benin, Ewè is an important language in West Africa. However, like many African languages, it remains underrepresented in natural language processing and speech recognition technologies. The Ewè audio classification challenge not only addresses a pressing social issue but also highlights the importance of linguistic diversity in AI development.
Challenge details:
– Dataset: Participants will receive access to a carefully curated dataset of Ewè audio samples.
– Task: Build models that can accurately classify Ewè audio clips into categories.
– Evaluation: Models will be judged based on classification accuracy.
Prizes:
- 1st place: $500
- 2nd place: $300
- 3rd place: $200
How to get involved:
1. Register: Visit the Zindi website to sign up and form your team.
2. Access resources: Download the Ewè audio dataset.
3. Develop: Build and train your audio classification models.
4. Submit: Upload your final model and results for evaluation before the closing date on the 29th of September.
5. Attend: Join us for the awards ceremony and knowledge-sharing sessions at the Moonshot by Techcabal conference, on the 9th of October.
Who should participate in the Ewè audio classification challenge?
We welcome participants from all backgrounds, including:
- Data scientists and machine learning engineers
- Linguists and language technology specialists
- Computer science students and researchers
- AI enthusiasts with a passion for language preservation
Why participate?
By participating in this challenge you will be;
- Harnessing the power of technology for social good.
- Contributing to the preservation and digitisation of an important African language.
- Gaining hands-on experience working with languages and audio data.
- Showcasing your skills to potential employers and collaborators in the AI industry.
- Networking with fellow participants and experts in the field.
The TechCabal Ewè audio classification challenge is a step towards a more inclusive digital future. The models and insights developed during this competition will facilitate the creation of language learning tools and support the evolution of speech recognition technology.
Join us on this exciting journey to harness the power of AI for linguistic diversity and cultural preservation.
