Low complexity techniques for embedded ASR systems

نویسندگان

  • Imre Kiss
  • Marcel Vasilache
چکیده

This paper deals with the problem of reducing the computational complexity of ASR algorithms for embedded systems. Particularly, three methods for simplifying the computation of state observation likelihoods of continuous density based HMMs are proposed. Feature component masking, variable-rate partial likelihood update and density pruning all result in significant savings in the decoding complexity with marginal impact on the recognition performance. A combination of feature component masking and density pruning was evaluated in a small vocabulary, 25lingual, speaker independent, isolated word recognition system. With a computational complexity reduction of 62% compared to the baseline system, a marginal, 1.6/6.5% relative error rate increase was obtained without/with on-line MAP adaptation on the average in clean and noisy operating environments. The presented framework can also be extended to larger vocabulary systems.

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

منابع مشابه

Towards large vocabulary ASR on embedded platforms

In this paper we present an overview of an automatic speech recognition system implementation in the context of embedded systems. Specific challenges presented by low resource platforms will be addressed for the basic components of an ASR decoder. Our main objective is to utilize and modify the technology developed for large vocabulary ASR to achieve efficient LVCSR on embedded systems as well.

متن کامل

Future study of Description System Architecture Approaches with Emphasis on Strategic Management

Systems Architecture is a generic discipline to handle objects (existing or to be created) called systems, in a way that supports reasoning about the structural properties of these objects. Systems Architecture is a response to the conceptual and practical difficulties of the description and the design of complex systems. Systems Architecture is a generic discipline to handle objects (existin...

متن کامل

Memory-Efficient Modeling and Search Techniques for Hardware ASR Decoders

This paper gives an overview of acoustic modeling and search techniques for low-power embedded ASR decoders. Our design decisions prioritize memory bandwidth, which is the main driver in system power consumption. We evaluate three acoustic modeling approaches–Gaussian mixture model (GMM), subspace GMM (SGMM) and deep neural network (DNN)–and identify tradeoffs between memory bandwidth and recog...

متن کامل

The Opportunities Afforded by Embedded Computer Systems for Monitoring and Control of Industrial Processes in Less-Industrialised Countries (TECHNICAL NOTE)

The dramatic changes in integrated-circuit technology over the last two decades have been of great benefit to countries such as Zimbabwe. High volume production of VLSI chips has produced a supply of intelligent, versatile electronic processing devices at very low cost. In particular the facilities of the microcontroller have steadily developed to the accompaniment of a reduction in price. Sinc...

متن کامل

Beginning of utterance detection algorithm for low complexity ASR engines

In this paper, a novel method for beginning of utterance detection is proposed for low complexity ASR systems. Assuming MFCC calculations in the ASR front-end, the additional computational load due to the algorithm is negligible. The algorithm makes use of the delay between the MFCC calculation and decoding process, which is typical in front-ends with feature normalization. The main steps of th...

متن کامل

ذخیره در منابع من


  با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

عنوان ژورنال:

دوره   شماره 

صفحات  -

تاریخ انتشار 2002