HN Showcase
Back to browse
Mimi in the browser – hear the semantic/acoustic split

Mimi in the browser – hear the semantic/acoustic split

by ymaws·Apr 20, 2026·4 points·1 comment
Visit ProjectView on HN

AI Analysis

●●SolidRabbit HoleWizardry

Hearing the semantic/acoustic split of neural audio codecs directly in your browser is wild.

Strengths
  • •Runs full ONNX audio encoding locally in-browser without server roundtrips.
  • •Interactive codebook toggling reveals how semantic and acoustic data separate.
  • •Clean UI makes complex transformer architecture intuitively understandable.
Weaknesses
  • •Tied entirely to Kyutai's Mimi model, not a general-purpose tool.
  • •Educational demo limits long-term utility beyond initial curiosity.
Category
AI/ML
Target Audience

ML engineers, audio developers, curious technologists

Similar To

Hugging Face Spaces · Kyutai Moshi · TensorFlow.js Audio

Similar Projects

Productivity●●Solid

Rhizome – semantic backlinks for your notes, generated locally

Writes wikilinks directly into files—no database, no daemon, just markdown.

Big BrainCozyNiche Gem
matzalazar
302mo ago
AI/ML●●●Banger

Decompose – Split text into classified semantic units, no LLM, 14ms

Non-LLM deterministic semantic decomposition—14ms, no hallucination, MCP-ready.

Big BrainSolve My ProblemWizardry
echology-io
104mo ago
AI/ML●●Solid

Instrumental Model from Scratch (With Demo)

The architecture is the project's real showpiece: a 72-band non‑uniform band-split BiMamba U‑Net that uses Mamba scans for O(T) memory and interleaved attention in the bottleneck to mix cross‑frequency context — a clever tradeoff between temporal efficiency and global attention. The author ships a runnable demo and an explanatory write-up so you can reproduce the approach, but it's clearly hobby-scale (≈1k songs trained, single home PC queue, slow cold starts), so expect experimental results rather than SOTA separation or instant throughput.

WizardryNiche Gem
day6
103mo ago