From the source
From the source
A collection of Pocket TTS models from 18.2M parameters down to 1.9M, and how we shrank the backbone, the sampling head and the decoder.
kyutai.org
From the source
From the source
From the source
A collection of Pocket TTS models from 18.2M parameters down to 1.9M, and how we shrank the backbone, the sampling head and the decoder.