Preprint
Attention-Guided Adaptation for Code-Switching Speech Recognition
Bobbi Aditya et al., arXiv, 2023
Read the originalRead on October 17, 2026
- What it is
- A 2023 method that adapts a multilingual Whisper model to Mandarin-English code-switching by training a small set of extra parameters. The abstract reports the resulting mixed error rate and the share of parameters trained.
- How much weight it carries
- A preprint evaluated on one corpus (SEAME) with Whisper-small. The baselines are the authors' own choices, so the numbers show what this adaptation achieved in this setup, not what other systems would reach.
Facts taken from this source
- A 2023 adaptation of Whisper reached a 14.2 percent mixed error rate on Mandarin-English code-switched speech, training 5.6 percent additional parameters.
“achieves a 14.2% mixed error rate, surpassing state-of-the-art method, where only 5.6% additional parameters over Whisper are trained” (Abstract)
- The authors say that multilingual models such as Whisper often struggle with the code-switching setting.
“these models often struggle with handling the code-switching setting” (Abstract)
Pages that rest on this source
- Multilingual Dictation: Two Languages, Auto-Detect, MixingGuide
- Can speech to text handle Hinglish, Spanglish and other mixed languages?Question
- Dictation Languages, Accents and Voice TranslationTopic
- Arabic Speech to Text and Right-to-Left DictationGuide
- Is speech recognition less accurate in languages other than English?Question