# Audio Scene Classification on HEAR Music 2021

Source: [Audio Scene Classification on HEAR Music 2021](<https://www.sota2.com/research/sota/audio-scene-classification-on-hear-music-2021>)

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

Data updated: 2026-02-26T05:37:43.049Z

Evaluation of Audio Scene Classification performance using the HEAR Music 2021 dataset, reporting results across various sub-tasks/datasets included in the suite.

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 | Audio Scene Classification |
| Dataset (including any stated split) | HEAR Music 2021 |
| Metric | Beijing |
| Unit | SCALAR |
| Direction | Higher is better |
| Results loaded | 5 of 5 |
| 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

- [Beijing](<https://www.sota2.com/research/sota/audio-scene-classification-on-hear-music-2021.md?metric=Beijing>): Higher is better; unit: SCALAR.

- [GTZAN-Genre](<https://www.sota2.com/research/sota/audio-scene-classification-on-hear-music-2021.md?metric=GTZAN-Genre>): Higher is better; unit: SCALAR.

- [GTZAN-M/S](<https://www.sota2.com/research/sota/audio-scene-classification-on-hear-music-2021.md?metric=GTZAN-M%2FS>): Higher is better; unit: SCALAR.

- [Mrd-Stroke](<https://www.sota2.com/research/sota/audio-scene-classification-on-hear-music-2021.md?metric=Mrd-Stroke>): Higher is better; unit: SCALAR.

- [Mrd-Tonic](<https://www.sota2.com/research/sota/audio-scene-classification-on-hear-music-2021.md?metric=Mrd-Tonic>): Higher is better; unit: SCALAR.

- [NSynth 5h](<https://www.sota2.com/research/sota/audio-scene-classification-on-hear-music-2021.md?metric=NSynth%205h>): Higher is better; unit: SCALAR.

- [NSynth 50h](<https://www.sota2.com/research/sota/audio-scene-classification-on-hear-music-2021.md?metric=NSynth%2050h>): Higher is better; unit: SCALAR.

- [Average Accuracy](<https://www.sota2.com/research/sota/audio-scene-classification-on-hear-music-2021.md?metric=Average%20Accuracy>): Higher is better; unit: SCALAR.

- [% Improvement over Random Baseline](<https://www.sota2.com/research/sota/audio-scene-classification-on-hear-music-2021.md?metric=%25%20Improvement%20over%20Random%20Baseline>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Beijing | Result date | Report paper |
| --- | --- | --- | --- |
| ABT AudioNTT | 0.966 | Not recorded | 68fa6e3e1e209131cea3ea1a |
| HEAR CREPE | 0.928 | Not recorded | 68fa6e3e1e209131cea3ea1a |
| HEAR BYOL-A\* | 0.919 | Not recorded | 68fa6e3e1e209131cea3ea1a |
| HEAR wav2vec2.0 | 0.907 | Not recorded | 68fa6e3e1e209131cea3ea1a |
| ABT ViTC-B (16 x 8) | 0.869 | Not recorded | 68fa6e3e1e209131cea3ea1a |

### ABT AudioNTT

Beijing: 0.966. Result ID: 699009993a6c314849fda796.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6e3e1e209131cea3ea1a.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6e3e1e209131cea3ea1a.md>)

[Implementation](<https://github.com/jonahanton/SSL_audio>)

Recorded evaluation configuration: {"Encoder":"AudioNTT"}. This record does not assert independent reproduction.

### HEAR CREPE

Beijing: 0.928. Result ID: 699009993a6c314849fda793.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6e3e1e209131cea3ea1a.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa64111e209131cea02831.md>)

Recorded evaluation configuration: {"Model":"CREPE"}. This record does not assert independent reproduction.

### HEAR BYOL-A\*

Beijing: 0.919. Result ID: 699009993a6c314849fda795.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6e3e1e209131cea3ea1a.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6dba1e209131cea3b9bf.md>)

[Implementation](<https://github.com/GasserElbanna/serab-byols>)

Recorded evaluation configuration: {"Model":"BYOL-A\*"}. This record does not assert independent reproduction.

### HEAR wav2vec2.0

Beijing: 0.907. Result ID: 699009993a6c314849fda794.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6e3e1e209131cea3ea1a.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa68891e209131cea1cc36.md>)

[Implementation](<https://github.com/pytorch/fairseq>)

Recorded evaluation configuration: {"Model":"wav2vec2.0"}. This record does not assert independent reproduction.

### ABT ViTC-B (16 x 8)

Beijing: 0.869. Result ID: 699009993a6c314849fda797.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6e3e1e209131cea3ea1a.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6e3e1e209131cea3ea1a.md>)

[Implementation](<https://github.com/jonahanton/SSL_audio>)

Recorded evaluation configuration: {"Encoder":"ViTC-B","Patch Size":"16x8"}. This record does not assert independent reproduction.

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