PTC-Bias: Phoneme-Level Temporal Competition for Bias Retrieval and Post-Decoding Correction in Speech LLMs

PTC-Bias：用于语音大模型偏置检索与解码后校正的音素级时间竞争框架

Abstract: Contextual biasing improves rare-word recognition in speech large language models (SpeechLLMs), but efficiently exploiting large bias lists remains challenging. We propose PTC-Bias, a two-stage framework based on phoneme-level temporal competition.

摘要： 上下文偏置（Contextual biasing）能够提升语音大模型（SpeechLLMs）对生僻词的识别能力，但如何高效利用大规模偏置词表仍是一项挑战。我们提出了 PTC-Bias，这是一个基于音素级时间竞争（phoneme-level temporal competition）的两阶段框架。

At the prefill stage, PTC Retrieval performs frame-synchronous phoneme decoding and temporal competition among candidate pronunciations, producing a compact bias-word shortlist and corresponding speech intervals.

在预填充（prefill）阶段，PTC 检索执行帧同步音素解码，并在候选发音之间进行时间竞争，从而生成一个精简的偏置词候选列表及对应的语音区间。

After SpeechLLM decoding, PTC Correction conducts a second local competition between the retrieved candidates and mismatched transcript spans within these intervals. Selective correction reduces near-homophone and word-segmentation errors while preserving correct transcriptions.

在语音大模型解码后，PTC 校正会在这些区间内，对检索到的候选词与转录中不匹配的片段进行第二次局部竞争。这种选择性校正减少了近音词和分词错误，同时保留了正确的转录内容。

Both stages share the same phoneme posteriors and require no additional SpeechLLM forward pass. Experiments on LibriSpeech show consistent gains across two SpeechLLMs and bias lists of up to 2000 words.

这两个阶段共享相同的音素后验概率，且无需额外的语音大模型前向传播。在 LibriSpeech 上的实验表明，该方法在两种语音大模型及高达 2000 个偏置词的词表下均表现出持续的性能提升。

With Prompt-SLAM-ASR-7B and 2000 bias words, PTC-Bias reduces B-WER by 23.4%/23.9% relative to CTC-Filter on test-clean/test-other, while keeping U-WER nearly unchanged.

在使用 Prompt-SLAM-ASR-7B 模型和 2000 个偏置词的情况下，与 CTC-Filter 相比，PTC-Bias 在 test-clean 和 test-other 数据集上的 B-WER（偏置词错误率）分别降低了 23.4% 和 23.9%，同时保持 U-WER（非偏置词错误率）几乎不变。