An approach for efficient open vocabulary spoken term detection

Norouzian, Atta; Rose, Richard C.

doi:10.1016/j.specom.2013.09.002

Cited by 8 publications

(9 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…While detection performance reported in [9] for IV queries was quite good, it was found that there was still a significant gap between the detection performance obtained for IV and OOV terms. The techniques presented in Section 2 significantly reduce this performance gap by using a more powerful strategy for verifying OOV term occurrences from lattice paths retrieved from the hybrid index.…”

Section: Introductionmentioning

confidence: 84%

“…Once the likelihoods of the paths containing the index term are scaled, all the paths of lj are re-ranked and then if the highest ranking path contains Vi that path is added to the index for Vi. The motivation and a detailed description of this indexing method can be found in [9].…”

Section: Offline -Building An Index Of Lattice Pathsmentioning

confidence: 99%

“…The performance of the second pass query term verification process is evaluated in terms of its ability to detect the locations of query terms in the retrieved audio segments. Additional discussion providing analysis of the word based lattice index construction is provided in [9].…”

Section: Experimental Studymentioning

confidence: 99%

“…The approach follows from previous work involving a hybrid indexing strategy for efficient open vocabulary STD [9]. The strategy, summarized in Section 2, relies on wordbased indexing terms, but uses a phoneme-based distance measure to associate the user's query with index entries.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Facilitating open vocabulary spoken term detection using a multiple pass hybrid search algorithm

Norouzian

Rose

2012

2012 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

Self Cite

View full text Add to dashboard Cite

This paper presents an efficient approach to spoken term detection (STD) from unstructured audio recordings using word lattices generated off-line from an automatic speech recognition (ASR) system.The approach facilitates open vocabulary STD and focuses specifically on reducing the difference between detection performance obtained for within-vocabulary (IV) and out-of-vocabulary (OOV) search terms. Improved OOV detection performance is obtained by using a two-pass search procedure. Candidate audio segments are retrieved from an index of word lattice paths in the first pass. Locations of OOV search terms are detected in the second pass from a constrained alignment of phonemic expansions of the query terms with phoneme sequences obtained from acoustic segments using an unconstrained neural network based phone decoder. It is found that the combination of first pass segment retrieval and second pass term verification significantly increases STD performance for OOV query terms with no increase in search time for utterances taken from a lecture speech domain.

show abstract

Section: Introductionmentioning

confidence: 84%

Section: Offline -Building An Index Of Lattice Pathsmentioning

confidence: 99%

Section: Experimental Studymentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Facilitating open vocabulary spoken term detection using a multiple pass hybrid search algorithm

Norouzian

Rose

2012

2012 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

Self Cite

View full text Add to dashboard Cite

show abstract

“…The most common approaches rely on lattices generated by a large vocabulary continuous speech recognition (LVCSR) system [1,2,3,4,5]. In addition, there are many alternative STD approaches in use that require fewer linguistic resources than those required by LVCSR systems [6,7].…”

Section: Introductionmentioning

confidence: 99%

Zero resource graph-based confidence estimation for open vocabulary spoken term detection

Norouzian

Rose

Ghalehjegh

et al. 2013

2013 IEEE International Conference on Acoustics, Speech and Signal Processing

Self Cite

View full text Add to dashboard Cite

In this paper the use of acoustic similarity of speech intervals for generating improved confidence scores for spoken term detection (STD) is investigated. A procedure based on acoustic dotplots which requires no training data is deployed for discovering similar speech intervals. A graph based random walk algorithm incorporates acoustic similarity of hypothesized term occurrences for improving the corresponding confidence scores. The proposed approach is evaluated in an open vocabulary STD task defined on a lecture domain corpus. It is shown that updating the confidence scores in this fashion results in a significant increase in term detection performance of out of vocabulary search terms. A relative improvement of 12.9% in figure of merit was gained relative to that obtained from a baseline lattice based STD system.

show abstract