AI 中文总结
Itgan系统利用LoRA适配Whisper，在NADI 2026三个阿拉伯语ASR子任务中分别通过方言专用模型、基础模型选择和权重平均等策略，取得57.1%、46.7%和14.49%的WER，并报告了无效方向。
AI 中文摘要
我们描述了Itgan系统在NADI 2026的三个ASR子任务上的表现，即稳健的国家级ASR（1.1）、混合方言ASR（1.2）和突尼斯代码切换ASR（1.3）。这三个子任务共享一个配方：在消费级GPU上使用LoRA适配Whisper，每个子任务通过不同的附加改进来实现。在1.1子任务中，测试时给定方言标签，从池化适配器继续训练的每种方言专用模型带来了最大的提升，提交的系统达到了57.1%的国家平均词错误率（WER）。在评估后，基于冻结编码器特征的线性探针可以在没有标签的情况下路由话语，并恢复了Oracle路由所提供性能的44%。在1.2子任务中，基础模型的选择比适配器容量更重要，系统组合只有在添加了一个去相关的成员后才有所帮助，达到了46.7%的WER。在1.3子任务中，我们的系统以14.49%的WER排名第二，并在领先的提交中拥有最低的字符错误率（CER），为5.38%。其最后的0.60个WER百分点是在没有进一步训练的情况下获得的，主要来自独立训练运行的精确权重空间平均，ROVER投票贡献了其余部分。每次比较都进行了配对自助法检验，我们报告了八个未奏效的方向。
英文摘要
We describe the Itgan systems for the three ASR subtasks of NADI 2026, namely robust country-level ASR (1.1), mixed-dialect ASR (1.2), and Tunisian code-switched ASR (1.3). All three share one recipe, Whisper adapted with LoRA on consumer GPUs, and each was carried by a different addition to it. On 1.1, where the dialect label is given at test time, per-dialect specialists continued from a pooled adapter gave the largest gain, and the submitted system reached 57.1% country-average WER. A post-evaluation linear probe on frozen encoder features routes utterances without the label and recovers 44% of what oracle routing gives. On 1.2 the choice of base model mattered more than adapter capacity, and system combination helped only once we added a decorrelated member, reaching 46.7% WER. On 1.3 our system placed second at 14.49% WER with the lowest CER among the leading submissions, 5.38%. Its last 0.60 WER points came without further training, mostly from an exact weight-space average of independently trained runs, with ROVER voting adding the remainder. Every comparison carries a paired-bootstrap test, and we report eight directions that did not work.
Comments12 pages, Arabic NLP 2026 Shared Task
