Voice dialogue model 'Haru 4B' capable of speaking naturally like a human is released for free
higurashi released the voice dialogue model 'Haru 4B', which supports natural human-like speech and interruptions, under the Apache License 2.0 for free.
Key points
- •The model enables natural human-like speech and handles backchannels and interruptions during conversation.
- •The pre-trained base model's open weights are available on Hugging Face under the Apache License 2.0.
- •Inference code is also publicly available on GitHub, allowing free usage.