Contaminated speech training methods for robust DNN-HMM distant speech recognition

By Mirco Ravanelli, Maurizio Omologo. 2017

32 citations recorded by the source corpus.

Despite the significant progress made in the last years, state-of-the-art speech recognition technologies provide a satisfactory performance only in the close-talking condition. Robustness of distant speech recognition in adverse acoustic conditions, on the other hand, remains a crucial open issue for future applications of human-machine interaction. To this end, several advances in speech enhancement, acoustic scene analysis as well as acoustic modeling, have recently contributed to improve the state-of-the-art in the field. One of the most effective approaches to derive a robust acoustic modeling is based on using contaminated speech, which proved helpful in reducing the acoustic mismatch between training and testing conditions. In this paper, we revise this classical approach in the context of modern DNN-HMM systems, and propose the adoption of three methods, namely, asymmetric context windowing, close-talk based supervision, and close-talk based pre-training. The experimental results, obtained using both real and simulated data, show a significant advantage in using these three methods, overall providing a 15% error rate reduction compared to the baseline systems. The same trend in performance is confirmed either using a high-quality training set of small size, and a large one.

Subfields: Artificial Intelligence, Safety, Robustness & Interpretability, Speech & Audio, Signal Processing

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Safety, Robustness & Interpretability

In Speech & Audio

In Signal Processing

Citations in the corpus

Multi-Task Self-Supervised Learning for Robust Speech Recognition (2020, positive): “Additive noises are extracted from FreeSound and DIRHA datasets [19, 20] and include both background and non-stationary noises such as alarms, door knocks, telephone ringing, television, and many others.”
The DIRHA-ENGLISH corpus and related tasks for distant-speech recognition in domestic environments (2015, background): “Our past experience [30, 33, 14] indeed suggests that, even though the use of language models is certainly helpful in increasing the recognition performance, the adoption of a simple phone-loop task is more suitable for experiments purely focusing on the acoustic information.”
Learning Speaker Representations with Mutual Information (2018, methodology): “In this case, all the clean signals were convoluted with a different impulse response [35], that was sampled from the DIRHA dataset [36, 37].”
The Pytorch-kaldi Speech Recognition Toolkit (2018, background): “For instance, it is possible to jointly load both context-dependent and context-independent targets and use the latter ones to perform monophone regularization [26, 27].”
Batch-normalized joint training for DNN-based distant speech recognition (2016, background): “Nevertheless, the recent success of deep learning has not only largely contributed to the substantial improvement of the speech recognition part of a DSR system [19, 20, 21, 22, 23, 24, 25], but has also enabled the de-a r X i v : 1703 .”
Interpretable Convolutional Filters with SincNet (2018, background): “Moreover, this work not only considers standard close-talking speech recognition, but it also extends the validation of SincNet to distant-talking speech recognition [54, 55, 56].”
A network of deep neural networks for Distant Speech Recognition (2017, background): “…recent success of deep learning has not only largely contributed to the substantial improvement of the speech recognition part of a DSR system [12–14], but has also enabled the development of competitive DNN-based speech enhancement solutions [15,16], making an effective integration be-tween…”
Realistic Multi-Microphone Data Simulation for Distant Speech Recognition (2016, background): “Approaches as contaminated speech training [14, 15, 16], multi-style training [17, 18, 19] and data augmentation [20, 21, 22, 23] have, in fact, been shown very effective in improving the DSR system performance.”
Attention Is All You Need In Speech Separation (2020, unknown): “Data contamination with noise or reverberation [28] is not used as well.”
Twin Regularization for online speech recognition (2018, background): “Acommon aspect of past approaches is that they often employ asymmetric context windows [22–24], that embed more past than future information.”