The Lincoln large-vocabulary HMM CSR

Abstract

The work described here focuses on recognition of the Wall Street Journal (WSJ) pilot database [17], a new CSR database which supports 5K, 20K, and up to 64K-word CSR tasks. The original Lincoln Tied-Mixture HMM CSR was implemented using a time-synchronous beam-pruned search of a static network [14] and does not extend well to this task because the recognition network would be too large for currently practical workstations. Therefore, the recognizer has been converted to a stack decoder-based search strategy [I,7,16]. This decoder has been shown to function effectively on up to 64K-word recognition of continuous speech. This paper describes the acoustic modeling techniques and the implementation of the stack decoder used to obtain these results.

DARPA February 1992 pilot corpus CSR "dry run" benchmark test results

“…[6], or "The tests.., reported here are little more than pilot tests for debugging purposes and no strong conclusions can be drawn." [7] and "Our strategy was to implement a system as quickly as possible in order to meet the tight CSR deadline. [8]"…”

“…[6], and "The results, however, show promise and will require more rigorous testing." [7] and "This is a preliminary report demonstrating that.., was ported from a 1000-word task (ATIS) to a large vocabulary task (5000-word) task.., in three man weeks." [8] Based on these observations, and on the experience gained in designing, collecting, and distributing the DARPA Pilot CSR Corpus, and in rapidly adapting existing technology to the new domain, there is good reason to look forward to the results of future benchmark tests.…”

“…The four sites (CMU [5], Dragon Systems [6], MIT/LL [7] and SRI [8]) provided NIST with a total of 22 sets of results for a number of test sets and system configurations. The number of test set results provided by individual contractors ranged from 1 to 10.…”