Datos personales

Mi foto
Fonoaudiólogas con amplia experiencia en manejo de población infantil, con trastornos del lenguaje, habla, alimentación y aprendizaje, atenderán a tus hijos promoviendo y fortaleciendo las habilidades necesarias para un adecuado desarrollo lingüístico y comunicativo. Contáctenos: hablapalabra@hablapalabra.com.co Tels: 3124204631 – 3114405956

viernes, 24 de septiembre de 2010

RELIABILITY OF THE PERCEPTUAL EVALUATION OF MP3 SPEECH SAMPLES


Author Block: Milena Cleves, SPEECH PATHOLOGIST1, Claudia Arboleda, SPEECH PATHOLOGIST2, Brigith Duenas, SPEECH PATHOLOGIST1, Karina Lizarraga, MD.3, Luis Bermudez, MD.4.

1Gilberto Marino Cleft Care Center, Bogota, Colombia, 2Operation Smile Boyaca Cleft Care Center, Bogota, Colombia, 3Operation Smile Inc., Norfolk, VA, USA, 4Operation Smile Inc., Bogota, Colombia.

ABSTRACT 

BACKGROUND & PURPOSE: Measuring functional outcomes of cleft palate surgery is a challenge. The perceptual evaluation of speech is currently the most available patient oriented method being used. The use of MP3 compressed audio files facilitates the managing and sharing of speech data between centers for blind evaluation. Most researchers consider MP3 files as non high fidelity samples and non suitable for reliable perceptual evaluation. It is the purpose of this study to develop an assessment tool (valid, reliable, available and easy to share) to measure the surgical outcomes of cleft palate repair and to test its reliability.


METHODS: Three Speech and Language Pathologists from two cleft care centers evaluated 106 speech samples taken in five different countries (Peru, Paraguay, Colombia, Honduras, Venezuela) .
Samples consisted of a set of conversational speech samples and five sentences (four containing oral consonants and one with a mixture of oral and nasal consonants).
The audio speech recordings were converted to MP3 format with a 128 kbts/sec bit rate. Samples were evaluated by each speech pathologist. The parameters described by Henningsoon G. et al (2008) were evaluated.


RESULTS: The result of inter-rater reliability tests were: hyper nasality (.51), hypo nasality (.94 ), voice disorder (.84 ),audible nasal emission and/or turbulence (.34 ), abnormal backing of oral target to post uvular place (.38), abnormal backing of oral targets but place remains oral (.28), nasal fricative (.21), nasal consonant for oral pressure consonant(.69), intelligibility (.48) and acceptability (.50).


CONCLUSIONS: This study has shown moderate to good inter-rater reliability on 6 of the 10 parameters evaluated. We think that the size of the speech sample was the cause of the fair reliability of the detailed consonant production errors. A new speech sample has been designed and it will be tested in the near future. The perceptual evaluation of MP3 compressed speech samples using the parameters proposed by Henningston et al, can be a reliable tool. It could be used to easily share the data between several centers or missions worldwide. It is recommended to take a longer sample with at least 10 sentences including all phonemes in several phonetic contexts to improve the reliability.

No hay comentarios:

Publicar un comentario