BA-SOT: 面向多说话人自动语音识别的边界感知序列化输出训练
BA-SOT: Boundary-Aware Serialized Output Training for Multi-Talker ASR
AI总结：
针对多说话人ASR中SOT方法说话人切换预测难的问题，提出BA-SOT方法，通过边界相关任务与损失、两阶段CTC策略优化，结合新指标UD-CER验证，显著降低了字符错误率。
摘要：
近期提出的序列化输出训练（SOT）通过生成由特殊 token 分隔的说话人文本，简化了多说话人自动语音识别（ASR）。但频繁的说话人切换会让说话人切换预测变得困难。为解决该问题，我们提出边界感知序列化输出训练（BA-SOT），它通过说话人切换检测任务和边界约束损失，将边界知识显式融入解码器。我们还引入了两阶段联结主义时间分类（CTC）策略，结合 token 级 SOT CTC 来恢复时序上下文信息。除了典型的字符错误率（CER），我们还引入了依赖话轮的字符错误率（UD-CER），以进一步衡量说话人切换预测的精度。与原始 SOT 相比，BA-SOT 将 CER/UD-CER 降低了 5.1%/14.0%；使用预训练 ASR 模型初始化 BA-SOT 模型，可进一步将 CER/UD-CER 降低 8.4%/19.9%。
英文摘要：
The recently proposed serialized output training (SOT) simplifies multi-talker automatic speech recognition (ASR) by generating speaker transcriptions separated by a special token. However, frequent speaker changes can make speaker change prediction difficult. To address this, we propose boundary-aware serialized output training (BA-SOT), which explicitly incorporates boundary knowledge into the decoder via a speaker change detection task and boundary constraint loss. We also introduce a two-stage connectionist temporal classification (CTC) strategy that incorporates token-level SOT CTC to restore temporal context information. Besides typical character error rate (CER), we introduce utterance-dependent character error rate (UD-CER) to further measure the precision of speaker change prediction. Compared to original SOT, BA-SOT reduces CER/UD-CER by 5.1%/14.0%, and leveraging a pre-trained ASR model for BA-SOT model initialization further reduces CER/UD-CER by 8.4%/19.9%.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
