LlamaVoice
A large speech generation model based on the Llama architecture.
CommonProductProgrammingSpeech GenerationMachine Learning
LlamaVoice is a large speech generation model built on the Llama architecture. It offers a more fluid and efficient processing approach by directly predicting continuous features, as opposed to the conventional vector quantization models that rely on discrete speech code prediction. The model includes key features such as continuous feature prediction, variational autoencoder (VAE) latent feature prediction, joint training, advanced sampling strategies, and flow-based enhancement.
LlamaVoice Visit Over Time
Monthly Visits
515580771
Bounce Rate
37.20%
Page per Visit
5.8
Visit Duration
00:06:42