R. Carr'e, R. Descout, M. Esk'enazi, J. Mariani, M. Rossi, "The French language database: defining, planning, and recording a large database," ICASSP-84.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....French. 2. EXPERIMENT The principle of the experiment is to let voluntary subject listen to some sentences more or less altered and to ask them to transcribe what they have understood. 2.1. Material We used a corpus of 50 phonetically balanced sentences [5] extracted from the database BD SONS [6, 7] (sentences groups PEQ01 to PEQ05) and uttered by 10 different speakers (5 females and 5 males) The average duration of the sentences was 3 seconds. The preparation of the stimuli has been done using MATLAB. It consists in reversing windows of n ms every n ms along a sentence (see Fig. 1) The ....

R. Carre, R. Descout, M. Eskenazi, J. Mariani, and M. Rossi, "The French language database: defining, planning, and recording a large database," in Proceedings of ICASSP 84, 1984, pp. 42.10.1--42.10.4.

....In this section we provide a brief description of the corpora used to carry out these experiments on identifying non linguistic speech features, and provide a baseline performance assessment for the phone recognizer. Five corpora have been used in the experiments reported in this paper: BDSONS[3] and BREF[24, 14] for French; TIMIT[9] and WSJ[35] for English, and the OGI 10language Corpus[32] BREF, TIMIT and WSJ0 have been used for sex identification; BREF and TIMIT for speaker identification; and all 5 corpora have been used for language identification. Since the training and test data ....

....BREF and TIMIT for speaker identification; and all 5 corpora have been used for language identification. Since the training and test data used differ for the various experiments, the details are specified later for each experiment. The BDSONS Corpus: BDSONS, Base de Donnees des Sons du Francais[3], was designed to provide a large corpus of French speech data for the study of the sounds in the French language and to aid speech research. The corpus contains an evaluation subcorpus consisting primarily of isolated and connected letters, digits and words from 32 speakers (16m 16f) and an ....

R. Carre, R. Descout, M. Eskenazi, J. Mariani, M. Rossi, "The French language database: defining, planning, and recording a large database," ICASSP-84.

....error rates as afunction of duration and language (with phonotactic constraints) Language identification accuracies are given in Table 1 ICASSP 94 with phonotactic constraints provided byphone bigrams. Results are given for 4 test corpora, WSJ[23] and TIMIT[4] for English, and BREF[6] and BDSONS[1] for French, as a function of the duration of the speech signal which includes approximately 100ms of silence. The initial and final silences were automatically removed based on HMM segmentation, so as to be able to compare language identification as a function of duration without biases due to ....

R. Carre , R. Descout, M. Eskenazi, J. Mariani, M. Rossi, "The French language database: defining, planning, and recording a large database," ICASSP-84.

....decoder is now implemented as a one pass beam search procedure applied on all the models in parallel, resulting in an efficient decoding procedure which saves a lot of computation. EXPERIMENTAL CONDITIONS Four corpora have been used to carry out the experiments reported in this paper: BDSONS[2] and BREF[15, 8] for French; and TIMIT[4] and WSJ0[22] for English. From the BDSONS corpus only the phonetically equilibrated sentence sub corpus (CDROM 6) has been used for testing, whereas depending on experiment, the 3 other corpora have been used for training and testing. The BDSONS Corpus: ....

....and WSJ0[22] for English. From the BDSONS corpus only the phonetically equilibrated sentence sub corpus (CDROM 6) has been used for testing, whereas depending on experiment, the 3 other corpora have been used for training and testing. The BDSONS Corpus: BDSONS, Base de Donnees des Sons du Francais[2], was designed to provide a large corpus of French speech data for the study of the sounds in the French language and to aid speech research. The corpus contains an evaluation subcorpus consisting primarily of isolated and connected letters, digits and words from 32 speakers (16m 16f) and an ....

R. Carre, R. Descout, M. Eskenazi, J. Mariani, M. Rossi, "The French language database: defining, planning, and recording a large database," ICASSP-84.

....BREF and TIMIT for speaker identification; and all 5 corpora have been used for language identification. Since the training and test data used differ for the various experiments, the details are specified later for each experiment. The BDSONS Corpus: BDSONS, Base de Donnees des Sons du Francais (Carre, Descout, Eskenazi, Mariani Rossi, 1984), was designed to provide a large corpus of French speech data for the study of the sounds in the French language and to aid speech research. The corpus contains an evaluation subcorpus consisting primarily of isolated and connected letters, digits and words from 32 speakers (16m 16f) and an ....

Carre, R., Descout, R., Eskenazi, M., Mariani, J., & Rossi, M. (1984). "The French language database: defining, planning, and recording a large database," Proceedings of the IEEE ICASSP-84, San Diego, CA, 42.10.1-4.

....data containing 7240 sentences from 84 speakers. Language identification accuracies are given in Table 11 with phonotactic constraints provided by a phone bigram. Language identification error rates are given for the 4 test corpora, WSJ and TIMIT for English[41, 11] and BREF and BDSONS for French[18, 30, 7], as a function of the duration of the speech signal. Approximately 100ms of silence are included at the beginning and end of each utterance (the initial and final silences were automatically removed based on HMM segmentation) so as to be able to compare language identification as a function of ....

R. Carr'e, R. Descout, M. Esk'enazi, J. Mariani, M. Rossi, "The French language database: defining, planning, and recording a large database," ICASSP-84.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC