# ARM-software/ML-KWS-for-MCU

Keyword spotting on Arm Cortex-M Microcontrollers

Repository: https://github.com/ARM-software/ML-KWS-for-MCU
Canonical: https://ross.abutalabs.com/products/ml-kws-for-mcu
Language: C
License: Apache-2.0
License Family: permissive
Topics: arm, machine-learning, python, cmsis-nn, deep-neural-networks, microcontrollers
Last push: 2019-04-10T15:17:43+00:00

## Health v2 (maintenance only)
Score: 32/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 0, release rhythm 35, longevity 100
- inputs: {"age_days": 3185, "days_push": 2702, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1249, forks 428 (observed 2026-08-28T04:04:07.948830+00:00)

## What it is
TensorFlow models and training scripts for keyword spotting (wake-word detection) on Arm Cortex-M microcontrollers, accompanying the 'Hello Edge' paper. It includes pretrained DNN/CNN/LSTM/GRU/CRNN/DS-CNN models, quantization guidance, and example deployment code for Cortex-M boards.

## Use cases
- train a keyword spotting model for microcontrollers
- deploy wake-word detection on Cortex-M
- quantize a neural network for embedded audio inference
- compare DNN CNN LSTM architectures for keyword spotting
- run speech command recognition on an MCU

## When to choose
- you need to build always-on voice wake-word detection on resource-constrained Arm Cortex-M hardware
- you want pretrained, quantized keyword spotting models with memory/accuracy tradeoffs from a research paper

## When to avoid
- you need modern maintained tooling - the repo is based on old TensorFlow 1.x and last released in 2019
- you need full speech recognition or large-vocabulary models rather than small keyword sets
- you target non-Arm or non-embedded platforms with modern frameworks like TensorFlow Lite Micro

## Facets
- artifact type: library
- maturity: maintenance
- function: machine-learning, deep-learning, audio-processing, speech-recognition, llm-training
- domain: machine-learning, deep-learning, embedded-systems, speech-processing
- platform: embedded, python, cpp, c, cross-platform
- tags: keyword-spotting, cortex-m, tensorflow, cmsis-nn, quantization, tinyml, wake-word, audio

## Member repositories
- ARM-software/ML-KWS-for-MCU (main) score 32

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:04:07.948830+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T05:07:45.957750+00:00, confidence not recorded.
  - readme: https://github.com/ARM-software/ML-KWS-for-MCU (fetched 2026-08-28T04:04:07.948830+00:00, sha 24e2e86d2245)
- Data as of 2026-08-30T08:39:29.467469+00:00.