Snips is joining Voice Lab, the first marketplace uniting French and European speech recognition technology players.
Today, the Big Tech companies and their Chinese counterparts dominate the booming speech recognition and artificial intelligence markets. But all that generated speech data is a major challenge for the future. There are of course the data security and confidentiality concerns, but this speech data is reaching a decisive volume, which is needed to train and enrich AI-based systems. To bridge the widening gap between digital giants and everyone else, it will be necessary to regroup, define new models, and offer alternatives. Voice Lab is doing just that.
Voice Lab: a marketplace uniting French and European players
Voice Lab is a French initiative bringing together diverse institutional players (universities, research laboratories) and private companies with a common interest in establishing an independent ecosystem and shared standards that enable France and Europe to remain competitive in the global voice tech market. Voice Lab is also set up to respond to strategic challenges posed by large groups and countries that are not compatible with American or Chinese solutions.
The first step will be to create a highly comprehensive French language corpus using available data and data input from industrial players. The impact of the French language is significant: French is the fifth largest language spoken in the world with 274 million speakers. This language data will be enhanced through automatic processing and then expanded, adding major European and international languages.
Voice Lab’s final objective is to define common API standards through a platform that enables fluid data collection, and to create a collective, neutral, and secure data storage environment that responds to the needs of many different players in European and international markets.
Large-scale data can challenge the virtual monopoly and technological advancement of Big Tech and Chinese counterparts.
Led by Karel Bourgeois, CEO of Voxist, this initiative already brings together nearly 30 players, including Voxist , Kwalys, and us here at Snips. COO Yann Lechelle believes this participation will generate positive results for the voice ecosystem:
“There is a need to level the playing field, starting with the French language, which is underrepresented when it comes to the availability of balanced spoken and tagged audio samples. The goal of the Voice Lab is to harvest up to 100,000 hours of such data to enable many more players to join the voice revolution without depending on the usual suspects.”
This week, Snips HQ in Paris will host the relaunch of the Smart Voice Meetup in partnership with Voice Lab. For more information about the meetup series, click here.