ESP32 voice assistant: AIMET AdaRound cuts wake-word model 8x without the 3.7% accuracy hit of naive 4-bit
carrycooldude · x · 2026-09-05
Developer @ashwinexe built a small voice assistant for the ESP32, inspired by a live on-device demo. Key details:
- Compressed the wake-word model 8x using Qualcomm's AIMET. Naive 4-bit rounding lost 3.7% accuracy; AIMET's AdaRound quantization recovered it.
- The wake word audio was generated with GradiumAI TTS in 22 different voices.
A practical reference for on-device quantization: naive 4-bit rounding genuinely hurts small models, and AdaRound-style methods are worth trying first.
More from Infra
- llama.cpp team launches Llama: one-click local AI on your Mac menu bar — Hugging Face · 2026-10-23
- China's daily token use hits 500T, up 4x as average price doubles to ¥13.46 — teortaxesTex · 2026-10-23
- New all-reduce kernels deliver 12% decode speedup for DeepSeek-V4.1 on PCIe-connected GPUs — AlpinDale · 2026-10-23
- Offshore oil rigs could be converted into AI data centers powered by stranded gas — clemnt · 2026-10-23
- Same native UI, two stacks: Swift build is ~80MB Mac-only, Zig build is ~10MB cross-platform — jedisct1 · 2026-10-23
- SurfSense: an open-source local NotebookLM with MCP agents and a prompt-caching lesson — FurtiveMirth · 2026-10-23