Peer-reviewed research
Whale: Large-Scale multilingual ASR model with w2v-BERT and E-Branchformer with large speech data
Yosuke Kashiwagi et al., arXiv (Interspeech 2025), 2025
Read the originalRead on October 17, 2026
- What it is
- A large multilingual speech model from Sony, evaluated on several benchmarks. Its Table 3 compares character error rate on the Corpus of Spontaneous Japanese (CSJ) with Whisper large-v3 and other open models.
- How much weight it carries
- Peer-reviewed conference paper. The company that built the model also reports the comparison, so the margin over Whisper is the authors' own result. The CSJ scores are for spontaneous Japanese lectures, not everyday dictation.
Facts taken from this source
- On the third evaluation set of the Corpus of Spontaneous Japanese (CSJ eval3), Whisper large-v3 had a character error rate of 16.5 percent in the paper's comparison table.
“Table 3: Comparison on CSJ data (CER %). size eval1 eval2 eval3 Whisper large-v3 1.55B 18.4 17.5 16.5” (Table 3, comparison on CSJ data)