Voice Concerns: The Need For Diversity In Voice AI Datasets (A September 2 Virtual Panel)
Door Mozilla | 24 augustus 2021
The latest installment of MozFest Dialogues & Debates will explore the exclusionary data sets that often power voice technology — and their repercussions.
Virtual assistants and voice-based interfaces are becoming more prevalent throughout society. But the data sets that power them frequently have blind spots, meaning certain communities and populations can’t access the technology’s full potential — or even access it at all.
Bias and exclusion have a long, shared history with technology. But this time the relationship is more pronounced: If Alexa isn’t familiar with your culture — accent, popular phrases, etc. — you simply don’t get to use Alexa. With voice-first interfaces rising in popularity, who are they leaving behind on the way up?
In this panel, Mozilla facilitates a conversation between researchers, artists, and technologists to unpack the repercussions of — and solutions to — exclusionary voice data.
This panel is slated for Thursday, September 2 at 11am ET, and titled “Voice Concerns: The Need For Diversity In Voice AI Datasets.”
Audience members are able to submit their questions ahead of time by tweeting @mozilla with the hashtag #DialoguesAndDebates.
Watch on Mozilla's Twitter and YouTube.
Our 40-minute panel will feature:
Hillary Juma | Hillary is Common Voice Community Manager at Mozilla. Mozilla Common Voice is an open-source initiative to make voice technology more inclusive. Contributors donate speech data to a public dataset, which anyone can then use to train voice-enabled technology.
Kola Tubosun | Kola is a writer, linguist and teacher who, in the past, has worked at Google as a speech linguist project manager and project manager for natural language processing. In 2016, Kola won the Premio Ostana International Award for Scriptures in the Mother Tongue — the only person of African descent to do so.
Kathleen Siminyu | Kathleen Siminyu is an AI researcher who has focused on Natural Language Processing for African languages. She works at Mozilla as a Machine Learning Fellow to support the development of a Kiswahili Common Voice dataset and to build speech transcription models for end use cases in the agricultural and financial domains.
Johann Diedrick | Johann is the director of engineering at Somewhere Good, a 2021 Mozilla Creative Media award recipient, and an adjunct professor at NYU's ITP program. Johann works on sonic engineering, machine listening, computational audio, and programmatic sound through A Quiet Life.
And moderator Xavier Harding | Xavier is a writer on the content team at Mozilla. Formerly, Xavier was a journalist covering consumer tech and the tech industry. In the past, Xavier’s written for Popular Science, BuzzFeed, Lifehacker, Mic, Newsweek, Fortune and Vox. He also hosts the Dialogues and Debates interview series, discussing tech topics like misinformation, contact-tracing and the role technology plays in addressing racial injustices.
MozFest Dialogues & Debates is Mozilla’s speaker series that usually occurs in-person each year. We’re expanding the series into virtual terrain, to bring internet users the information they need to understand and advocate for a healthy and humane digital world. Watch past MozFest Dialogues & Debates here.