Viewing profile — watsonmusic

it's not oss

bonus usage

11labs is facing a real competitor

genius

this model is superb

Microsoft is cool

yes the best

one of the best models built by Microsoft

https://github.com/microsoft/VibeVoice

https://huggingface.co/microsoft/VibeVoice-1.5B

VibeVoice is a novel framework designed for generating expressive, long-form, multi-speaker conversational audio, such as podcasts, from text. It addresses significant challenges i…

cannot wait seeing how it goes beyond the current llm training pipeline

it could be adaptive. only high-value tokens were allocated with more compute

A new scaling paradigm finally comes out!

14b model performs comparably with 32b size. the improvement is huge

[dead]

[dead]

[dead]

[dead]

negative values can enhance the expressibility

not all hallucinations are creativity Imaginate that for a RAG application, the model is supposed to follow the given documents

the model is supposed to learn this

[flagged]

The modification is simple and beautiful. And the improvements are quite significant.