# Keyword Spotting on Speech Commands

Source: [Keyword Spotting on Speech Commands](<https://www.sota2.com/research/sota/keyword-spotting-on-speech-commands>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-03-16T10:15:17.061Z

Evaluation of Keyword Spotting performance using the Speech Commands dataset, measuring accuracy alongside hardware efficiency metrics like latency, power, and energy consumption.

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

| Scope | Recorded value |
| --- | --- |
| Task | Keyword Spotting |
| Dataset (including any stated split) | Speech Commands |
| Metric | Accuracy |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 4 of 4 |
| Coverage | All results currently returned for this benchmark |

Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol.

## Metrics

- [Accuracy](<https://www.sota2.com/research/sota/keyword-spotting-on-speech-commands.md?metric=Accuracy>): Higher is better; unit: PERCENT.

- [Latency (ms)](<https://www.sota2.com/research/sota/keyword-spotting-on-speech-commands.md?metric=Latency%20(ms)>): Lower is better; unit: TIME.

- [Power (mW)](<https://www.sota2.com/research/sota/keyword-spotting-on-speech-commands.md?metric=Power%20(mW)>): Lower is better; unit: OTHER.

- [Energy/Inf (mJ)](<https://www.sota2.com/research/sota/keyword-spotting-on-speech-commands.md?metric=Energy%2FInf%20(mJ)>): Lower is better; unit: OTHER.

- [Efficiency (GOp/s/W)](<https://www.sota2.com/research/sota/keyword-spotting-on-speech-commands.md?metric=Efficiency%20(GOp%2Fs%2FW)>): Higher is better; unit: OTHER.

## Recorded results

| Method | Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| MobileNetV2 | 94.7 | Not recorded | 69818200a8b01e76cc8e2a04 |
| NeuEdge | 93.2 | Not recorded | 69818200a8b01e76cc8e2a04 |
| Standard SNN | 91.4 | Not recorded | 69818200a8b01e76cc8e2a04 |
| This work | 77 | Not recorded | 69b78e7a0636e5bf4afbdbc1 |

### MobileNetV2

Accuracy: 94.7. Result ID: 6999ac295698665ec423b488.

[Reporting paper](<https://www.sota2.com/research/paper/69818200a8b01e76cc8e2a04.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa65811e209131cea0a9a1.md>)

[Implementation](<https://github.com/tensorflow/models/tree/master/research/slim/nets/mobilenet>)

Recorded evaluation configuration: {"Platform":"Jetson Nano"}. This record does not assert independent reproduction.

### NeuEdge

Accuracy: 93.2. Result ID: 6999ac295698665ec423b48a.

[Reporting paper](<https://www.sota2.com/research/paper/69818200a8b01e76cc8e2a04.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69818200a8b01e76cc8e2a04.md>)

Recorded evaluation configuration: {"Platform":"TrueNorth"}. This record does not assert independent reproduction.

### Standard SNN

Accuracy: 91.4. Result ID: 6999ac295698665ec423b489.

[Reporting paper](<https://www.sota2.com/research/paper/69818200a8b01e76cc8e2a04.md>)

Recorded evaluation configuration: {"Platform":"TrueNorth"}. This record does not assert independent reproduction.

### This work

Accuracy: 77. Result ID: 69b7d83247f34ce924a024a3.

[Reporting paper](<https://www.sota2.com/research/paper/69b78e7a0636e5bf4afbdbc1.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69b78e7a0636e5bf4afbdbc1.md>)

Recorded evaluation configuration: {"Hardware":"Physical","Number of nodes":"2–4"}. This record does not assert independent reproduction.

Page 1 of 1. This is an index page, not the full collection.