NIESR: Nuisance Invariant End-to-end Speech Recognition
This addresses robustness in speech recognition for applications like voice assistants by reducing overfitting to nuisances without costly labeled data, though it is incremental as it builds on existing invariance methods.
The paper tackles the problem of speech recognition models overfitting to nuisance factors like speaker identity and background noise by proposing an unsupervised adversarial invariance induction framework that separates essential factors from nuisances without extra labels. The method achieves relative improvements of 5.48% on WSJ0, 6.16% on CHiME3, and 6.61% on TIMIT over the base model, with a 14.44% improvement on a combined dataset.
Deep neural network models for speech recognition have achieved great success recently, but they can learn incorrect associations between the target and nuisance factors of speech (e.g., speaker identities, background noise, etc.), which can lead to overfitting. While several methods have been proposed to tackle this problem, existing methods incorporate additional information about nuisance factors during training to develop invariant models. However, enumeration of all possible nuisance factors in speech data and the collection of their annotations is difficult and expensive. We present a robust training scheme for end-to-end speech recognition that adopts an unsupervised adversarial invariance induction framework to separate out essential factors for speech-recognition from nuisances without using any supplementary labels besides the transcriptions. Experiments show that the speech recognition model trained with the proposed training scheme achieves relative improvements of 5.48% on WSJ0, 6.16% on CHiME3, and 6.61% on TIMIT dataset over the base model. Additionally, the proposed method achieves a relative improvement of 14.44% on the combined WSJ0+CHiME3 dataset.