Spoken Language Identification with Pre-trained Models and Margin Loss
For researchers in spoken language identification, this work provides an incremental improvement by applying known techniques (pre-trained models and margin losses) to a specific challenge dataset.
The paper tackles speaker-controlled spoken language identification by proposing a method using a pre-trained ECAPA-TDNN with margin-based losses, achieving 85.95% macro accuracy and 90.96% micro accuracy on the Tidy-X dataset, with 45.7% and 15.2% improvements over the baseline, respectively.
For the speaker-controlled spoken language identification task proposed in the TidyLang Challenge 2026, this paper proposes a language identification method based on pre-trained models and margin-based losses. The proposed method adopts a pre-trained ECAPA-TDNN as the feature encoder and incorporates margin-based losses to enhance the discriminative ability of language representations, thereby improving inter-class separability and reducing the interference of non-linguistic factors such as speaker characteristics. Experimental results on the Tidy-X dataset show that the proposed method achieves 85.95% macro accuracy and 90.96% micro accuracy on the language identification task and 17.08% equal error rate (EER) on the verification task. Compared with the official baseline, the macro accuracy improves by 45.7%, the micro accuracy improves by 15.2%, and the EER is reduced by approximately 50.8%, demonstrating the effectiveness of the proposed method. The code will be released at https://github.com/PunkMale/TidyLang2026.