面向音频大语言模型中英双语码转换语音识别的直接偏好优化
Direct Preference Optimization for English-Mandarin Code-Switching Speech Recognition in Audio LLMs
AI总结:
针对音频大语言模型在英中码转换语音转录中的系统失败,提出使用直接偏好优化(DPO)对齐模型,通过构建偏好对(保留混合语言内容 vs 模仿失败模式)训练模型,实现词错误率降低最高89.6%(分布内)和20.0%(分布外)。
摘要:
音频大语言模型(Audio LLMs)尽管具有强大的多语言能力,但在转录码转换语音时表现出系统性失败。聚焦英中双语,我们识别出三种失败模式:语言省略、翻译替代转录和幻觉。我们应用直接偏好优化(DPO)来对齐模型,构建偏好对,其中选择响应保留混合语言内容,而拒绝响应模仿失败模式。在100K对(570小时)上训练三个Audio LLMs,我们观察到一致的行为转变:模型学会在提示转录时保留语言组成而非翻译。这种对齐使得词错误率降低高达89.6%(分布内)和20.0%(分布外)。我们的发现表明,DPO可以有效地从多语言Audio LLMs中引出正确的码转换转录行为。
英文摘要:
Audio large language models (Audio LLMs) exhibit systematic failures in transcribing code-switching speech despite strong multilingual capabilities. Focusing on English-Mandarin, we identify three failure modes: language omission, translation-instead-of-transcription, and hallucination. We apply Direct Preference Optimization (DPO) to align models, constructing preference pairs in which chosen responses preserve mixed-language content while rejected responses mimic failure patterns. Training three Audio LLMs on 100K pairs (570 hours), we observe consistent behavioral shifts: models learn to preserve language composition rather than translating when prompted for transcription. This alignment yields MER reductions up to 89.6% (in-distribution) and 20.0% (out-of-distribution). Our findings suggest DPO can effectively elicit correct code-switching transcription behavior from multilingual Audio LLMs.
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。