Claim Missing Document
Check
Articles

Found 2 Documents
Search

Speech Recognition for Cleft Lip and Palate Voice and Standard Voice for Consonant Words /B/ Anggoro, Akhmad; Herdjunanto, Samiadji; Hidayat, Risanuri
SENATIK STT Adisutjipto Vol 5 (2019): Peran Teknologi untuk Revitalisasi Bandara dan Transportasi Udara [ISBN XXX-XXX-XXXXX-
Publisher : Sekolah Tinggi Teknologi Adisutjipto

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.28989/senatik.v5i0.325

Abstract

Advances in technology make speech recognition improve. But does speech recognition recognize the sounds of cleft lip and palate? this research uses the voice of cleft lip and palate to normal voice. With the letter / b / which is the letter for lip articulation. Words used include Clothes, Ash and Moist. The extraction method uses Mel Frequency Cepstral Coefficients (MFCC), the classification uses K-Nearest Neighbor (KNN) with K-Fold Cross-Validation as a test. The results show the accuracy above 70%. 75% in the word "Baju". 75% in the word "Abu". 83% in the word "Lembab".
MFCC DAN KNN UNTUK PENGENALAN SUARA ARTIKULASI P Anggoro, Akhmad; Herdjunanto, Samiadji; Hidayat, Risanuri
AVITEC Vol 2, No 1 (2020): Februari 2020
Publisher : Sekolah Tinggi Teknologi Adisutjipto

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.28989/avitec.v2i1.605

Abstract

Cleft lip and palate (CLP) is a term for patients who experience speech organ disorders, that disorder is caused by a gap found in the lip or palate. Patients will experience speech problems. Pattern recognition in CLP sound is still small in Indonesia. In this research in the language identification of CLP and standard sound patterns using the extraction of the Mel Frequency Cepstral Coefficients (MFCC) feature with K-Nearest Neighbor (KNN) classification and K-Fold cross-validation. By making words that have the letter /p/ as a reference, known as bilabial. The words used include Paku, Kapak, and Atap. The accuracy of recognition results reached more than 69%, with a minimum accuracy of 41%.