tensorjournal.com · Sat 14 Nov

Saturday 14 November

tiny-voca: A 4.5M-parameter text-to-speech model that runs entirely in your browser

AI-generated summary Developer nato_nob has showcased tiny-voca, a text-to-speech model with 4,537,291 parameters that runs locally within web browsers. The model uses a source-filter synthesizer instead of a neural vocoder and was trained on synthetic data generated by Qwen3-TTS.

Also covered byTensor Journal

Open Vision Banana outscores Vision Banana on Cityscapes, its authors report

A drafting table holds a line drawing of a city street beside a matching sheet where the same street is filled with flat cobalt, mint and off-white colour regions.

AI-generated summary The authors of Open Vision Banana report that their fully open recipe for generative visual perception scores 73.9 mIoU on Cityscapes zero-shot compared to 69.9 mIoU for Vision Banana. The paper submitted to arXiv by Arpit Jadon and co-authors details fine-tuning FLUX.2-Klein base generators at 4B and 9B scale on 2.1M-3.8M pairs.