A method to generate speech across multiple speakers
Role in this project:
ML Engineer Contributions:18 commits, 1 PR, 14 pushes in 1 year 5 months
Contributions summary:Adam primarily contributed to the development and refinement of a speech generation system. Their work included making the Visdom visualization optional, improving phoneme extraction, and adding support for single-speaker models, specifically the Blizzard 2011 model. Further contributions involved addressing dependencies, refactoring code to remove dependencies during the generation process and fixing masked loss, demonstrating a focus on optimization and modularity.
methodspeech-recognitionspeakersspeechtts
Contributions:154 commits in 1 year 3 months