The Voice Infrastructure for African Languages

All Voices is a community-driven platform for collecting and validating speech data across African languages. It enables native speakers, researchers, and institutions to contribute high-quality voice recordings that help train AI systems like Mansa.

Why this matters?

AI cannot learn languages it has never heard

Many African languages lack the datasets required to build reliable language technologies. By recording speech, transcribing audio, and validating samples, contributors help create the data needed to train systems such as:

Speech Recognition

Transcription Systems

Translation Tools

Conversational AI

Download All Voices

WHAT ALL VOICES DOES

What All Voices Does

helps organizations build speech and Text datasets

All Voices enables contributors to collect and validate speech and text data for African languages, creating datasets for AI training. Key features:

Structured Data Collection: Record speech, translate text, transcribe audio, and add speaker or dialect info.
Language Tasks: Transcribe speech and translate text to generate additional data.
Validation & Quality: Review and rate contributions to ensure accuracy

Validated data to supports speech recognition, translation, and transcription systems.