Parakeet Redux: is a 178 MB speech model useful in the real world?

Description and use

<p>Moondream recently released <a href="https://huggingface.co/moondream/parakeet-redux" rel="noopener noreferrer">Parakeet Redux</a>, a speech-to-text model that fits in 178 MB. It's NVIDIA's <a href="https://huggingface.co/nvidia/parakeet-tdt-0.6b-v3" rel="noopener noreferrer">Parakeet TDT 0.6B v3</a>, one of the strongest open speech models, with its weights cut down to three values each. The original is 1,256 MB, and the model card says accuracy stays close to it.</p> <p>A 0.6B-parameter model in roughly the space of a 61M one would change where good speech recognition can run: on an ordin

Problem

<p>Moondream recently released <a href="https://huggingface.co/moondream/parakeet-redux" rel="noopener noreferrer">Parakeet Redux</a>, a speech-to-text model that fits in 178 MB. It's NVIDIA's <a href="https://huggingface.co/nvidia/parakeet-tdt-0.6b-v3" rel="noopener noreferrer">Parakeet TDT 0.6B v3</a>, one of the strongest open speech models, with its weights cut down to three values each. The original is 1,256 MB, and the model card says accuracy stays close to it.</p> <p>A 0.6B-parameter model in roughly the space of a 61M one would change where good speech recognition can run: on an ordin

Outcome

<p>Moondream recently released <a href="https://huggingface.co/moondream/parakeet-redux" rel="noopener noreferrer">Parakeet Redux</a>, a speech-to-text model that fits in 178 MB. It's NVIDIA's <a href="https://huggingface.co/nvidia/parakeet-tdt-0.6b-v3" rel="noopener noreferrer">Parakeet TDT 0.6B v3</a>, one of the strongest open speech models, with its weights cut down to three values each. The original is 1,256 MB, and the model card says accuracy stays close to it.</p> <p>A 0.6B-parameter model in roughly the space of a 61M one would change where good speech recognition can run: on an ordin

Evidence

Parakeet Redux: is a 178 MB speech model useful in the real world?

Source