A Novel Loss Function and Training Strategy for Noise-Robust Keyword Spotting

Iván López Espejo, Zheng-Hua Tan, Jesper Jensen

Research output: Contribution to journalJournal articleResearchpeer-review

13 Citations (Scopus)
81 Downloads (Pure)


The development of keyword spotting (KWS) systems that are accurate in noisy conditions remains a challenge. Towards this goal, in this paper we propose a novel training strategy relying on multi-condition training for noise-robust KWS. By this strategy, we think of the state-of-the-art KWS models as the composition of a keyword embedding extractor and a linear classifier that are successively trained. To train the keyword embedding extractor, we also propose a new (C_{N,2}+1)-pair loss function extending the concept behind related loss functions like triplet and N-pair losses to reach larger inter-class and smaller intra-class variation. Experimental results on a noisy version of the Google Speech Commands Dataset show that our proposal achieves around 12% KWS accuracy relative improvement with respect to standard end-to-end multi-condition training when speech is distorted by unseen noises. This performance improvement is achieved without increasing the computational complexity of the KWS model.

Original languageEnglish
Article number9465680
JournalIEEE/ACM Transactions on Audio, Speech, and Language Processing
Pages (from-to)2254 - 2266
Number of pages13
Publication statusPublished - Jul 2021


  • Keyword spotting
  • deep metric learning
  • keyword embedding
  • loss function
  • multi-condition training
  • noise robustness


Dive into the research topics of 'A Novel Loss Function and Training Strategy for Noise-Robust Keyword Spotting'. Together they form a unique fingerprint.

Cite this