Abstract
In this work we propose the early incorporation of confidence information in the decoding process of large vocabulary speech recognition. A confidence based pruning technique is used to guide the search to the most promising paths. We introduce a posterior probability-based confidence measure that can be estimated efficiently and synchronously from the available information during the search process. The accuracy of this measure is enhanced using a discriminative training technique whose objective is to maximize the discrimination between the correct and incorrect decoding hypotheses. For this purpose, phone-level confidence scores are combined to derive word level scores. Highly compact models that exhibit minimal degradation in performance are introduced. Experimental results using large speech corpora show that the proposed method improves both the decoding accuracy and the decoding time when compared to a baseline recognition system that uses a conventional search approach. Furthermore, the introduced confidence measures are well-suited for cross-task portability.
| Original language | English |
|---|---|
| Pages (from-to) | 409-428 |
| Number of pages | 20 |
| Journal | Speech Communication |
| Volume | 42 |
| Issue number | 3-4 |
| DOIs | |
| State | Published - Apr 1 2004 |
Fingerprint
Keywords
- Confidence measure
- Discriminative training
- Pruning
- Speech recognition
ASJC Scopus subject areas
- Signal Processing
- Electrical and Electronic Engineering
- Experimental and Cognitive Psychology
- Linguistics and Language
Cite this
Beam search pruning in speech recognition using a posterior probability-based confidence measure. / Abdou, Sherif; Scordilis, Michael S.
In: Speech Communication, Vol. 42, No. 3-4, 01.04.2004, p. 409-428.Research output: Contribution to journal › Article
}
TY - JOUR
T1 - Beam search pruning in speech recognition using a posterior probability-based confidence measure
AU - Abdou, Sherif
AU - Scordilis, Michael S
PY - 2004/4/1
Y1 - 2004/4/1
N2 - In this work we propose the early incorporation of confidence information in the decoding process of large vocabulary speech recognition. A confidence based pruning technique is used to guide the search to the most promising paths. We introduce a posterior probability-based confidence measure that can be estimated efficiently and synchronously from the available information during the search process. The accuracy of this measure is enhanced using a discriminative training technique whose objective is to maximize the discrimination between the correct and incorrect decoding hypotheses. For this purpose, phone-level confidence scores are combined to derive word level scores. Highly compact models that exhibit minimal degradation in performance are introduced. Experimental results using large speech corpora show that the proposed method improves both the decoding accuracy and the decoding time when compared to a baseline recognition system that uses a conventional search approach. Furthermore, the introduced confidence measures are well-suited for cross-task portability.
AB - In this work we propose the early incorporation of confidence information in the decoding process of large vocabulary speech recognition. A confidence based pruning technique is used to guide the search to the most promising paths. We introduce a posterior probability-based confidence measure that can be estimated efficiently and synchronously from the available information during the search process. The accuracy of this measure is enhanced using a discriminative training technique whose objective is to maximize the discrimination between the correct and incorrect decoding hypotheses. For this purpose, phone-level confidence scores are combined to derive word level scores. Highly compact models that exhibit minimal degradation in performance are introduced. Experimental results using large speech corpora show that the proposed method improves both the decoding accuracy and the decoding time when compared to a baseline recognition system that uses a conventional search approach. Furthermore, the introduced confidence measures are well-suited for cross-task portability.
KW - Confidence measure
KW - Discriminative training
KW - Pruning
KW - Speech recognition
UR - http://www.scopus.com/inward/record.url?scp=1842475644&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=1842475644&partnerID=8YFLogxK
U2 - 10.1016/j.specom.2003.11.002
DO - 10.1016/j.specom.2003.11.002
M3 - Article
AN - SCOPUS:1842475644
VL - 42
SP - 409
EP - 428
JO - Speech Communication
JF - Speech Communication
SN - 0167-6393
IS - 3-4
ER -