Accurate client-server based speech recognition keeping personal data on the client

Georges, Munir; Kanthak, Stephan; Klakow, Dietrich

doi:10.1109/icassp.2014.6854205

Cited by 4 publications

(4 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This hybrid ASR system computes its final recognition result using server-side ASR system combination, in which the recognition result from a highly personalized on-device ASR system is combined with the results from a more generic server based ASR system. Similar hybrid approaches to ASR personalization have been explored before [22,23,24], albeit not to our knowledge in the context of at scale FT of the personalized on-device ASR system.…”

Section: Asr Personalizationmentioning

confidence: 99%

Federated Evaluation and Tuning for On-Device Personalization: System Design & Applications

Paulik¹,

Seigel²,

Mason³

et al. 2021

Preprint

View full text Add to dashboard Cite

We describe the design of our federated task processing system. Originally, the system was created to support two specific federated tasks: evaluation and tuning of on-device ML systems, primarily for the purpose of personalizing these systems. In recent years, support for an additional federated task has been added: federated learning (FL) of deep neural networks. To our knowledge, only one other system has been described in literature that supports FL at scale. We include comparisons to that system to help discuss design decisions and attached trade-offs. Finally, we describe two specific large scale personalization use cases in detail to showcase the applicability of federated tuning to on-device personalization and to highlight application specific solutions.

show abstract

Section: Asr Personalizationmentioning

confidence: 99%

Federated Evaluation and Tuning for On-Device Personalization: System Design & Applications

Paulik¹,

Seigel²,

Mason³

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

“…However, data privacy regulations as well as increasing demand for cloud offloading motivates execution on low power platforms without any need for cloud connection as proposed by, e.g., Georg Stemmer et al [31]. It is also in discussion to distribute the computation between embedded devices such as mobile phones and the cloud as proposed, e.g., by Munir Georges et al [32] or Alice Coucke et al [33]. Recent development goes in the direction of end-to-end and all neuronal network systems.…”

Section: Intent Classificationmentioning

confidence: 99%

Ultra-Compact NLU: Neuronal Network Binarization as Regularization

Georges

Czarnowski²

2019

Interspeech 2019

Self Cite

View full text Add to dashboard Cite

This paper describes an approach for intent classification and tagging on embedded devices, such as smart watches. We describe a technique to train neuronal networks where the final neuronal network weights are binary. This enables memory bandwidth optimized inference and efficient computation even on constrained/embedded platforms. The flow of the approach is as follows: tf-idf word selection method reduces the number of overall weights. Bag-of-Words features are used with a feedforward and recurrent neuronal network for intent classification and tagging, respectively. A novel double Gaussian based regularization term is used to train the network. Finally, the weights are almost clipped lossless to −1 or 1 which results in a tiny binary neuronal network for intent classification and tagging. Our technique is evaluated using a text corpus of transcribed and annotated voice queries. The test domain is "lights control". We compare the intent and tagging accuracy of the ultra-compact binary neuronal network with our baseline system. The novel approach yields comparable accuracy but reduces the model size by a factor of 16: from 160kB to 10kB.

show abstract

“…Examples are: improved protection of user privacy, lower recognition latency, or the autonomy from communications channels. Local inference has been previously addressed for speech recognition [8] and spoken language understanding [9]. The main challenges in local inference is the limited compute and memory available on the device.…”

Section: Introductionmentioning

confidence: 99%

Compact Speaker Embedding: lrx-Vector

Georges

Huang

2020

Interspeech 2020

Self Cite

View full text Add to dashboard Cite

Deep neural networks (DNN) have recently been widely used in speaker recognition systems, achieving state-of-the-art performance on various benchmarks. The x-vector architecture is especially popular in this research community, due to its excellent performance and manageable computational complexity. In this paper, we present the lrx-vector system, which is the low-rank factorized version of the x-vector embedding network. The primary objective of this topology is to further reduce the memory requirement of the speaker recognition system. We discuss the deployment of knowledge distillation for training the lrx-vector system and compare against low-rank factorization with SVD. On the VOiCES 2019 far-field corpus we were able to reduce the weights by 28% compared to the full-rank x-vector system while keeping the recognition rate constant (1.83 % EER).

show abstract

Accurate client-server based speech recognition keeping personal data on the client

Cited by 4 publications

References 16 publications

Federated Evaluation and Tuning for On-Device Personalization: System Design & Applications

Federated Evaluation and Tuning for On-Device Personalization: System Design & Applications

Ultra-Compact NLU: Neuronal Network Binarization as Regularization

Compact Speaker Embedding: lrx-Vector

Contact Info

Product

Resources

About