Turkic Wikimedia Conference 2023/Submissions/Language, Speech, Mozilla Common Voice and Voice AI

ID :	Language, Speech, Mozilla Common Voice and Voice-AI
Author(s): Bülent Özden		Username(s): Bulentozden	Type of submission: 1 hour (2 sessions?) lecture
		Affiliation: Mozilla Common Voice Turkish Language Representative	Theme(s): Culture, Education, Technology, Community Engagement, AI
Abstract: The lecture introduces common concepts and issues of culture, language, and speech in relation to voice-AI with a computational linguistics perspective. The voice-AI model generation workflow will be covered in general. The open-source (CC0) Mozilla Common Voice project, which aims to democratize voice-AI, will be introduced in detail. Some model generation results and possible applications will be presented. The lecture will also cover common concepts like responsive AI, bias, and cultural/economic/environmental aspects of AI. It will also represent the problems we faced during our fair-use attempt at Turkish Wikipedia and its language quality, and how we solved them. Turkic languages in the Common Voice Project & voice-AI arena will be used as examples.			Slides: N/A at the moment
			Level of advancement: basic & medium
			Special requirements: I need to use my own laptop and need an audio output connection to the sound system.
How will this session be beneficial for the communities in the region of Turkic world? Although one of the most major language groups, Turkic languages have been under constant pressure and many of them died in time, and some are still endangered. Wikipedia and Common Voice projects are both very valuable to preserve the culture related to communities speaking these languages, in written and spoken form. Common Voice and Voice-AI will help us remedy some of these problems. The lecture will not be too technical, but will also include some aspects which will help Wikipedians working on AI.

Interested participants

(register below and ask your questions now to the session organizer)

• ...