用多步预测优化稀有词识别，显著降低语音识别错误率
Efficient Trie-based Biasing using K-step Prediction for Rare Word Recognition
- 通过提前预测多步输出，避免传统方法中的分数回撤步骤
- 在NSC Part 2测试集上，错误率从30.86%降至12.19%
- 仅需10小时合成数据微调Whisper模型，高效实用
上下文偏置通过在解码过程中优先输出稀有词来提升语音识别模型对稀有词的识别效果。常见方法是基于Trie的偏置，为可能导向稀有词（如“Bonham”）的部分假设（如“Bon”）赋予额外分数。若最终未生成完整稀有词，则系统会撤销这些加分。此机制受限于束搜索，且计算开销大，尤其对大型解码器模型。为此，我们提出让ASR模型具备前瞻能力，一次性预测多个步骤。该方法通过更准确估计部分假设是否通向完整稀有词，彻底避免了分数回撤。仅用10小时合成数据微调Whisper模型，在NSC Part 2测试集上将词错误率从30.86%降至12.19%。
原文摘要 · Abstract (English)
Contextual biasing improves rare word recognition of ASR models by prioritizing the output of rare words during decoding. A common approach is Trie-based biasing, which gives "bonus scores" to partial hypothesis (e.g. "Bon") that may lead to the generation of the rare word (e.g. "Bonham"). If the full word ("Bonham") isn't ultimately recognized, the system revokes those earlier bonuses. This revocation is limited to beam search and is computationally expensive, particularly for models with large decoders. To overcome these limitations, we propose adapting ASR models to look ahead and predict multiple steps at once. This avoids the revocation step entirely by better estimating whether a partial hypothesis will lead to the generation of the full rare word. By fine-tuning Whisper with only 10 hours of synthetic data, our method reduces the word error rate on the NSC Part 2 test set from 30.86% to 12.19%.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。