Sentence examples for representation of the query from inspiring English sources

The first level relies on an optimized representation of the query as a cascade of phonetic filters.

When performing 3D shape retrieval, the CBW representation of the query shape is constructed on line, and compared with those in the database.

Amphion accepts user queries formulated with the help of a graphical query-formulation guide, producing a diagrammatic representation of the query.

Next to mapping query text to concepts using string matching, a common method to obtain a MeSH-based representation of the query is to use relevance feedback: The original query is used to retrieve a set of documents and based on the MeSH terms assigned to these documents, a MeSH-based query is obtained.

This system employs a Mel-frequency cepstral coefficient (MFCC) fingerprint representation of the queries and utterances [67], and search is performed using a dynamic time warping approach.

All were based on a feature representation of the queries and the utterances and a DTW-based search to hypothesize detections.

For feature extraction, the phoneme recognizers developed at BUT for Czech, English, Hungarian, and Russian languages [76] are employed to obtain a posteriorgram-based representation of the queries and the utterances.

This system consists of the fusion of three different subsystems, as shown in Fig. 1: a large vocabulary continuous speech recognition system, a dynamic time warping search-based system with a fingerprint representation of the queries and the utterances, and a DTW search-based system with phoneme posterior probabilities for query and utterance representation.

This system, whose architecture is shown in Fig. 3, consists of the fusion of two different subsystems: the Kaldi LVCSR-based system and the DTW search-based system with an MFCC fingerprint representation of the queries and utterances, both described previously.

As a baseline, we only use the textual representation of the queries and documents.

There are approximately 63 k different phoneme trigrams in the final index, and the number of items exceeds 88 M. When searching the word index, all possible phonetic transcriptions (phoneme representations) of the query word are found in the lexicon.