Samsung Research at SpeechBrain 2023 Online Summit - ISCA 2023 Interspeech Satellite Event

The Samsung AI Center – Cambridge (SAIC Cambridge), is proud to contribute to helping SpeechBrain in maintaining and evolving his code base. With this official Interspeech 2023 Satellite Event, SAIC-Cambridge takes one step further and teams up with academics from the Montreal Institute of Learning Algorithms (MILA, Montreal, Quebec, Canada) and Avignon University (France) to foster for the very first time since the raise of deep-learning-based open-source toolkits a large discussion within the speech community. The event will take place four days after the main Interspeech conference on August 28th.

In just a few words, SpeechBrain has established itself as a leading deep learning toolkit for speech processing. With an average of 100 daily clones and 1000 daily downloads on its GitHub repository, along with over 6,000 stars and 1100 forks, SpeechBrain is a popular choice among speech processing experts. SAIC-Cambridge latest contributions to the toolkit include, but are not limited to, recipes for state-of-the-art speech recognition on the Librispeech dataset with Conformer models, and significant efforts towards helping in releasing the next version of SpeechBrain, that we will present in details during this event.

With this Summit, we hope to initiate discussions within speech community, facilitate networking, and draw a clear plan for the future of open-source speech technologies including SpeechBrain, but also other toolkits such as ESPNet, K2, HuggingFace etc. Of course, SAIC-Cambridge and the other organizers plan to deliver all the latest news about the toolkit to the audience, and we hope to get as much feedbacks as possible. Open-source toolkits exist thanks to the users.

Invited talks will include speakers from JP Morgan Chase & Co, ViaDialog, UbenwaAI, Orange Labs as well as the University of Cambridge and the University of Avignon. The day will be concluded with a long panel discussion grouping major actors coming from the open-source community: Kaldi, ESPnet, Librosa, HuggingFace and Torchaudio.

If you are interested, visit our official website and register yourself, it’s free!: