Pocketsphinx: A Free, Real-Time Continuous Speech Recognition System for Hand-Held Devices

Abstract

The availability of real-time continuous speech recognition on mobile and embedded devices has opened up a wide range of research opportunities in human-computer interactive applications. Unfortunately, most of the work in this area to date has been confined to proprietary software, or has focused on limited domains with constrained grammars. In this paper, we present a preliminary case study on the porting and optimization of CMU SPHINX-II, a popular open source large vocabulary continuous speech recognition (LVCSR) system, to hand-held devices. The resulting system operates in an average 0.87 times real-time on a 206MHz device, 8.03 times faster than the baseline system. To our knowledge, this is the first hand-held LVCSR system available under an open-source license.

A comparative review of dynamic neural networks and hidden Markov model methods for mobile on-device speech recognition

“…These techniques include, but are not limited to, maximum likelihood estimation [7] and a semi-continuous approach, where the unexpected transitions are managed with the introduction of a language transition map [8]. The Sphinx II speech recognition system uses the latter approach and the Pocketsphinx speech recognition system, developed for use on some mobile devices, is an adaptation of this system [9].…”

Evaluation of Methods to Combine Different Speech Recognizers

“…After that, all recognizers produces output. Output of the recognizer is the hypothesis: score of how well audio signal matches the acoustic model [8]. This hypothesis score is passed to classification algorithm and it makes final decision.…”

“…It is distributed under the same permissive license as Sphinx toolkit itself. Algorithmically this is hidden Markov model based speech recognition framework, which provides simple way for creating custom speech recognition systems [8].…”

VOICECONET: A Collaborative Framework for Speech-Based Computer Accessibility with a Case Study for Brazilian Portuguese

“…Julius [24] and Sphinx-4 [51] are examples of open source decoders that can be used in association with the required resources to build ASR engines. For embedded (e.g., smartphones) applications, Julius and PocketSphinx [19] are the most popular engines. Research groups have made available engines for English, Japanese and other languages based on open source decoders Silva et al [40].…”