System and method for dynamic noise adaptation for robust automatic speech recognition
View Patent ↗A speech processing method and arrangement are described. A dynamic noise adaptation (DNA) model characterizes a speech input reflecting effects of background noise. A null noise DNA model characterizes the speech input based on reflecting a null noise mismatch condition. A DNA interaction model performs Bayesian model selection and re-weighting of the DNA model and the null noise DNA model to realize a modified DNA model characterizing the speech input for automatic speech recognition and compensating for noise to a varying degree depending on relative probabilities of the DNA model and the null noise DNA model.
1. A method comprising:
characterizing, by a computing device, a speech input based on a dynamic noise adaptation (DNA) model reflecting effects of background noise;
characterizing the speech input based on a null noise DNA model reflecting a null noise mismatch condition; and
performing Bayesian model selection and re-weighting of the DNA model and the null noise DNA model to realize a modified DNA model characterizing the speech input for automatic speech recognition and compensating for noise to a varying degree depending on relative probabilities of the DNA model and the null noise DNA model,
wherein the Bayesian model selection and re-weighting reflects a competing likelihood of which model best characterizes the speech input, and
wherein re-weighting the DNA model and the null noise DNA model includes assigning zero probability to the DNA model predicted by Bayesian model averaging when the DNA model does not best characterize the speech input.
2. The method of claim 1 , wherein re-weighting the DNA model and the null noise DNA model includes averaging.
3. The method of claim 1 , wherein re-weighting the DNA model and the null noise DNA model includes increasing the probability of the DNA model predicted by Bayesian model averaging when the DNA model best characterizes the speech input.
4. The method of claim 1 , wherein the DNA model includes a probability based noise model reflecting transient and evolving components of a current noise estimate.
5. A non-transitory computer-readable medium storing computer-readable instructions that, when executed by a processor, cause a device to:
characterize a speech input based on a dynamic noise adaptation (DNA) model reflecting effects of background noise;
characterize the speech input based on a null noise DNA model reflecting a null noise mismatch condition; and
perform Bayesian model selection and re-weighting of the DNA model and the null noise DNA model to realize a modified DNA model characterizing the speech input for automatic speech recognition and compensating for noise to a varying degree depending on relative probabilities of the DNA model and the null noise DNA model,
wherein the Bayesian model selection and re-weighting reflects a competing likelihood of which model best characterizes the speech input, and
wherein re-weighting the DNA model and the null noise DNA model includes assigning zero probability to the DNA model predicted by Bayesian model averaging when the DNA model does not best characterize the speech input.
6. The non-transitory computer-readable medium of claim 5 , wherein re-weighting the DNA model and the null noise DNA model includes averaging.
7. The non-transitory computer-readable medium of claim 5 , wherein re-weighting the DNA model and the null noise DNA model includes increasing the probability of the DNA model predicted by Bayesian model averaging when the DNA model best characterizes the speech input.
8. The non-transitory computer-readable medium of claim 5 , wherein the DNA model includes a probability based noise model reflecting transient and evolving components of a current noise estimate.
9. A method comprising:
characterizing, by a computing device, a speech input based on a dynamic noise adaptation (DNA) model reflecting effects of background noise;
characterizing the speech input based on a null noise DNA model reflecting a null noise mismatch condition; and
performing Bayesian model selection and re-weighting of the DNA model and the null noise DNA model to realize a modified DNA model characterizing the speech input for automatic speech recognition and compensating for noise to a varying degree depending on relative probabilities of the DNA model and the null noise DNA model,
wherein the Bayesian model selection and re-weighting reflects a competing likelihood of which model best characterizes the speech input,
wherein re-weighting the DNA model and the null noise DNA model includes reducing the probability of the DNA model predicted by Bayesian model averaging when the DNA model does not best characterize the speech input, and
wherein re-weighting the DNA model and the null noise DNA model includes increasing the probability of the DNA model predicted by Bayesian model averaging when the DNA model best characterizes the speech input.
10. The method of claim 9 , wherein re-weighting the DNA model and the null noise DNA model includes averaging.
11. The method of claim 9 , wherein re-weighting the DNA model and the null noise DNA model includes assigning zero probability to the DNA model predicted by Bayesian model averaging when the DNA model does not best characterize the speech input.
12. The method of claim 9 , wherein the DNA model includes a probability based noise model reflecting transient and evolving components of a current noise estimate.