Papers › Triplet Entropy Loss: Improving The Generalisation of Short Speech Language...

Triplet Entropy Loss: Improving The Generalisation of Short Speech Language Identification Systems

3 Dec 2020arXiv:2012.03775archive 2025-07-28

Ruan van der Merwe

We present several methods to improve the generalisation of language identification (LID) systems to new speakers and to new domains. These methods involve Spectral augmentation, where spectrograms are masked in the frequency or time bands during training and CNN architectures that are pre-trained on the Imagenet dataset. The paper also introduces the novel Triplet Entropy Loss training method, which involves training a network simultaneously using Cross Entropy and Triplet loss. It was found that all three methods improved the generalisation of the models, though not significantly. Even though the models trained using Triplet Entropy Loss showed a better understanding of the languages and higher accuracies, it appears as though the models still memorise word patterns present in the spectrograms rather than learning the finer nuances of a language. The research shows that Triplet Entropy Loss has great potential and should be investigated further, not only in language identification tasks but any classification task.

PaperPDFCode

Code

ruanvdmerwe/triplet-entropy-loss officialmentioned on GitHubtf report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

Not run by Syntology. Nothing on this page verifies that the listed code works.

Tasks

Language IdentificationSpeech Language IdentificationSpoken Language UnderstandingSpoken language identification

1 archive task tag without a task page not shown.

Results from the paper archive 2025-07-28

No leaderboard rows for this paper in the archive.

Methods

Introduced by this paper: Triplet Entropy Loss

1x1 ConvolutionAdamAverage PoolingBatch NormalizationBottleneck Residual BlockConcatenated Skip ConnectionConvolutionDense BlockDense ConnectionsDropoutGlobal Average PoolingKaiming InitializationMax PoolingReLUResidual BlockResidual ConnectionSoftmaxTriplet Entropy Loss

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections