Keyword Spotting with African Languages

Keyword spotting refers to the task of learning to detect spoken keywords. It interfaces all modern voice-based virtual assistants on the market: Amazon’s Alexa, Apple’s Siri, and the Google Home device. Contrarily to speech recognition models, keyword spotting doesn’t run on the cloud, but directly on the device. 

The motivation of this paper is to extend the Speech commands dataset (Warden 2018) with African languages. In particular, we are going to focus on 6 Senegalese languages: Wolof, Pulaar, Serer, Mandinka, Diola, Soninke. 

The choice of these languages is guided, on the one hand, by their status as languages considered to be the languages of the first generation, that is to say, the first codified languages (endowed with a writing system and considered by the state of Senegal as national languages) with decree n ° 68-871 of July 24, 1968. On the other hand, they represent the languages that are most spoken in Senegal.