WAXAL: A large-scale open resource for African language speech technology

2026-03-06 · Source: The latest research from Google · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Data Science & Analytics · Depth: Intermediate, medium

Summary

WAXAL is a new, large-scale, open-access speech dataset released by Google Research on March 6, 2026, designed to support African language speech technology. It covers 27 Sub-Saharan African languages spoken by over 100 million people across more than 26 countries. The dataset includes approximately 1,846 hours of transcribed natural speech for Automatic Speech Recognition (ASR) and over 565 hours of high-fidelity recordings for Text-to-Speech (TTS). Released under a Creative Commons CC-BY-4.0 license, WAXAL aims to bridge the digital divide by providing crucial data for low-resource languages, enabling the development of robust, inclusive voice-enabled technologies tailored to Africa's linguistic diversity. The project involved multi-year collaboration with African academic and community organizations.

Key takeaway

For AI Engineers and NLP Engineers developing speech technologies for diverse linguistic populations, WAXAL offers a critical, permissively licensed resource. Your teams can leverage this dataset to build more accurate and inclusive ASR and TTS systems for 27 Sub-Saharan African languages, directly addressing the data scarcity challenge. Consider integrating WAXAL to expand your models' linguistic coverage and improve performance in low-resource contexts.

Key insights

WAXAL provides a large, open-access dataset for 27 African languages, fostering inclusive speech technology development.

Principles

• Open access accelerates research.
• Community collaboration ensures relevance.
• Natural speech data improves ASR.

Method

WAXAL-ASR uses image-prompted elicitation for natural, unscripted speech. WAXAL-TTS employs collaborative script drafting and studio recordings for high-fidelity audio.

In practice

• Use WAXAL for ASR model training.
• Utilize WAXAL-TTS for synthetic voice generation.
• Explore image-prompted data collection.

Topics

• African Language Speech Technology
• Automatic Speech Recognition
• Text-to-Speech
• Open-Access Datasets
• Low-Resource Languages

Code references

• openai/whisper
• google-research/bert

Best for: AI Engineer, NLP Engineer, AI Scientist, AI Researcher, Machine Learning Engineer, Data Scientist

Related on AIssential

• BaltiVoice: A Speech Corpus and Fine-tuned Whisper ASR System for the Balti Language — Fine-tuning pre-trained ASR models with modest, newly created corpora can establish functional speech recognition for low-resource languages.
• Paza: Introducing automatic speech recognition benchmarks and models for low resource languages — Human-centered design and evaluation are crucial for effective speech models in low-resource, real-world environments.
• AfriVoices-KE: A Multilingual Speech Dataset for Kenyan Languages — AfriVoices-KE provides a large, diverse speech dataset for five underrepresented Kenyan languages.
• Granite 4.0 1B Speech: Compact, Multilingual, and Built for the Edge — Granite 4.0 1B Speech offers high accuracy and multilingual support in a compact, efficient model for edge devices.
• Software for Automatic Speech Recognition via Whisper models applied to Oral History interviews in the Portuguese language — Ethos AT provides private, accessible automatic speech transcription for Portuguese using Whisper models.
• Adding Benchmaxxer Repellant to the Open ASR Leaderboard — Private datasets enhance ASR leaderboard trustworthiness by preventing benchmaxxing and revealing nuanced model performance.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by The latest research from Google.