# Contextual Automatic Speech Recognition on ContextASR-Bench

Source: [Contextual Automatic Speech Recognition on ContextASR-Bench](<https://www.sota2.com/research/sota/contextual-automatic-speech-recognition-on-contextasr-bench>)

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

Data updated: 2026-03-27T16:49:03.729Z

Evaluation of Contextual Automatic Speech Recognition performance on the ContextASR-Bench dataset, measured using various Word Error Rate (WER) and Named Entity Error Rate (NE-WER) metrics across English and Mandarin speech and dialogue contexts.

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 | Contextual Automatic Speech Recognition |
| Dataset (including any stated split) | ContextASR-Bench |
| Metric | Average Error Rate |
| Unit | PERCENT |
| Direction | Lower is better |
| Results loaded | 6 of 6 |
| 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

- [Average Error Rate](<https://www.sota2.com/research/sota/contextual-automatic-speech-recognition-on-contextasr-bench.md?metric=Average%20Error%20Rate>): Lower is better; unit: PERCENT.

- [WER (Speech, English)](<https://www.sota2.com/research/sota/contextual-automatic-speech-recognition-on-contextasr-bench.md?metric=WER%20(Speech%2C%20English)>): Lower is better; unit: PERCENT.

- [NE-WER (Speech, English)](<https://www.sota2.com/research/sota/contextual-automatic-speech-recognition-on-contextasr-bench.md?metric=NE-WER%20(Speech%2C%20English)>): Lower is better; unit: PERCENT.

- [NE-FNR (Speech, English)](<https://www.sota2.com/research/sota/contextual-automatic-speech-recognition-on-contextasr-bench.md?metric=NE-FNR%20(Speech%2C%20English)>): Lower is better; unit: PERCENT.

- [WER (Dialogue, English)](<https://www.sota2.com/research/sota/contextual-automatic-speech-recognition-on-contextasr-bench.md?metric=WER%20(Dialogue%2C%20English)>): Lower is better; unit: PERCENT.

- [NE-WER (Dialogue, English)](<https://www.sota2.com/research/sota/contextual-automatic-speech-recognition-on-contextasr-bench.md?metric=NE-WER%20(Dialogue%2C%20English)>): Lower is better; unit: PERCENT.

- [NE-FNR (Dialogue, English)](<https://www.sota2.com/research/sota/contextual-automatic-speech-recognition-on-contextasr-bench.md?metric=NE-FNR%20(Dialogue%2C%20English)>): Lower is better; unit: PERCENT.

- [WER (Speech, Mandarin)](<https://www.sota2.com/research/sota/contextual-automatic-speech-recognition-on-contextasr-bench.md?metric=WER%20(Speech%2C%20Mandarin)>): Lower is better; unit: PERCENT.

- [NE-WER (Speech, Mandarin)](<https://www.sota2.com/research/sota/contextual-automatic-speech-recognition-on-contextasr-bench.md?metric=NE-WER%20(Speech%2C%20Mandarin)>): Lower is better; unit: PERCENT.

- [NE-FNR (Speech, Mandarin)](<https://www.sota2.com/research/sota/contextual-automatic-speech-recognition-on-contextasr-bench.md?metric=NE-FNR%20(Speech%2C%20Mandarin)>): Lower is better; unit: PERCENT.

- [WER (Dialogue, Mandarin)](<https://www.sota2.com/research/sota/contextual-automatic-speech-recognition-on-contextasr-bench.md?metric=WER%20(Dialogue%2C%20Mandarin)>): Lower is better; unit: PERCENT.

- [NE-WER (Dialogue, Mandarin)](<https://www.sota2.com/research/sota/contextual-automatic-speech-recognition-on-contextasr-bench.md?metric=NE-WER%20(Dialogue%2C%20Mandarin)>): Lower is better; unit: PERCENT.

- [NE-FNR (Dialogue, Mandarin)](<https://www.sota2.com/research/sota/contextual-automatic-speech-recognition-on-contextasr-bench.md?metric=NE-FNR%20(Dialogue%2C%20Mandarin)>): Lower is better; unit: PERCENT.

## Recorded results

| Method | Average Error Rate | Result date | Report paper |
| --- | --- | --- | --- |
| Ming-Flash-Omni | 3.3 | Not recorded | 6902f0f23ef7ccd74e5e4274 |
| Qwen3-ASR | 4.23 | Not recorded | 6902f0f23ef7ccd74e5e4274 |
| Qwen3-Omni-30B-A3B-Instruct | 4.34 | Not recorded | 6902f0f23ef7ccd74e5e4274 |
| Gemini2.5-Pro | 6.3 | Not recorded | 6902f0f23ef7ccd74e5e4274 |
| Baichuan-Omni-1.5 | 8.17 | Not recorded | 6902f0f23ef7ccd74e5e4274 |
| Kimi-Audio | 9.24 | Not recorded | 6902f0f23ef7ccd74e5e4274 |

### Ming-Flash-Omni

Average Error Rate: 3.3. Result ID: 69c6b48c47f34ce924b119f2.

[Reporting paper](<https://www.sota2.com/research/paper/6902f0f23ef7ccd74e5e4274.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6902f0f23ef7ccd74e5e4274.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### Qwen3-ASR

Average Error Rate: 4.23. Result ID: 69c6b48c47f34ce924b119f0.

[Reporting paper](<https://www.sota2.com/research/paper/6902f0f23ef7ccd74e5e4274.md>)

[Method origin paper](<https://www.sota2.com/research/paper/697c3afda8b01e76cc8da424.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### Qwen3-Omni-30B-A3B-Instruct

Average Error Rate: 4.34. Result ID: 69c6b48c47f34ce924b119f1.

[Reporting paper](<https://www.sota2.com/research/paper/6902f0f23ef7ccd74e5e4274.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7b881e209131cea88f3d.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### Gemini2.5-Pro

Average Error Rate: 6.3. Result ID: 69c6b48c47f34ce924b119ed.

[Reporting paper](<https://www.sota2.com/research/paper/6902f0f23ef7ccd74e5e4274.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7c311e209131cea8cef5.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### Baichuan-Omni-1.5

Average Error Rate: 8.17. Result ID: 69c6b48c47f34ce924b119ef.

[Reporting paper](<https://www.sota2.com/research/paper/6902f0f23ef7ccd74e5e4274.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### Kimi-Audio

Average Error Rate: 9.24. Result ID: 69c6b48c47f34ce924b119ee.

[Reporting paper](<https://www.sota2.com/research/paper/6902f0f23ef7ccd74e5e4274.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78431e209131cea77803.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

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