List your company

ResearchFormerly WizWand

Speech Recognition

Benchmarks

DatasetSOTA methodMetricTrendResultsLast updated

Original Audio 176 utterances (Combined)

11.85WER

5

Feb 26, 2026

Original Audio Noisy

19.06WER

5

Feb 26, 2026

Original Audio Clean 94 utterances

6.56WER

5

Feb 26, 2026

Librispeech (100 hours) and AISHELL v1

5.4WER (English)

5

Feb 26, 2026

Vaani 48 languages (test)

27.4WER

4

Jun 24, 2026

UIED Out-of-Domain

6.4WER

4

May 26, 2026

ST-AEDS (Out-of-Domain)

3.7WER

4

May 26, 2026

EmoV-DB (Out-of-Domain)

16.8WER

4

May 26, 2026

EMNS (Out-of-Domain)

7.4WER

4

May 26, 2026

MedDictate English (test)

4.6WER

4

May 19, 2026

SAP 5-fold CV (test)

Context-cond. LoRA

6.59WER

4

May 5, 2026

Vibration Sensor Audio Mobile Scenarios, Inside lab

7CER

4

Mar 10, 2026

General Throughput Evaluation

Whisper Baseline

168.9Throughput (tokens/s)

4

Mar 9, 2026

14.43WER

4

Mar 9, 2026

Google Speech Commands (GSC)

95.6Accuracy

4

Feb 27, 2026

Google Voice Search (test)

6.2WER

4

Feb 26, 2026

3.61WER

4

Feb 26, 2026

SMS-2 eval92 (test)

18.9WER

4

Feb 26, 2026

SMS-2 (cv_dev93)

16.6WER

4

Feb 26, 2026

Simulated FLEURS (Wearer: English, Partner: Italian) (test)

13.3Wearer WER [%]

4

Feb 26, 2026

Simulated FLEURS (Wearer: English, Partner: French) (test)

12.9Wearer WER (%)

4

Feb 26, 2026

Simulated FLEURS (Wearer: English, Partner: Spanish) (test)

12.5Wearer WER

4

Feb 26, 2026

FLEURS 34 languages (test)

8.2WER

4

Feb 26, 2026

Long-form video (test)

2.08Avg WER

4

Feb 26, 2026

BABEL Lao lo (test)

29.7WER

4

Feb 26, 2026