Project

Validation of the Acoustic Voice Quality Index Version 03.01 and the Acoustic Breathiness Index in American English

Goal: Problem Statement and Impact: Instrumental voice assessment is a process that considers various methods of data collection, such as endoscopic assessment, aerodynamic assessment, self-assessment, auditory perceptual assessment, and objective acoustic voice analysis (Dejonckere et al., 2001; Patel et al., 2018). Measurement and assessment of the acoustic parameters of the voice have been one of the most commonly used approaches by voice clinicians as it is easily accessible and inexpensive to implement (Behlau & Murry, 2012; Mehta & Hillman, 2008). Currently, one of the dilemmas associated with acoustic voice analysis relates to the acoustic parameters of voice that one wishes to measure, which many of them underlie a particular type of voice sample, for example, the use of sustained vowel versus continuous speech tasks.

Background: Voice acoustic parameters come in many forms depending on the goal of the parameter, source of the acoustic signal, and the technique of analysis. Some parameters are more primary to voice production (e.g., fundamental frequency, loudness level), others are more related to voice quality (e.g., spectral slope, alpha ratio, CPP), others still are related to short term transient production effects (e.g., jitter, shimmer, relative fundamental frequency). Many of the voice acoustic parameters are obtained from the fundamental frequency (f0) of a sustained vowel, limiting the validity and reliability of acoustic measurements, so such measures should be interpreted with caution (Mehta & Hillman, 2008). Also, analysis of continuous speech samples has been shown to have more robust correlations with perceptual voice analysis (Lowell, 2012). However, it has also been postulated that speech samples alone may not accurately quantify what they are intended to measure due to the variability associated with speech characteristics, so their construct validity has also been questioned (Gerratt et al., 2016). Nevertheless, evidence has shown that software that allows concatenated analysis of audio samples, i.e., both continuous speech and a sustained vowel, provides reliable results concerning diagnostic accuracy and concurrent validity in overall voice quality when compared to auditory perceptual voice analysis (Barsties & De Bodt, 2015), the gold standard for assessing the severity of dysphonia (Maryn & Roy, 2012; Oates, 2009).

Overall Objective: The general aim of this study is to study the benefits and limitations of acoustic parameters when considering both steady production and running speech towards a more robust and valid assessment protocol that voice clinicians and voice scientists alike could use.

Current Goal: The specific goal of this study is further test the validity of a multidimensional analysis of voice called Acoustic Voice Quality Index (AVQI) Version 03.01 and the Acoustic Breathiness Index (ABI), which considers for its final result a multiple regression analysis of six concatenated acoustic variables, i.e., obtained from both a sample of sustained vowel and speech and thus generate a single score that predicts the severity of dysphonia (Batthyany et al., 2019; Maryn et al., 2010). The most recent version of the AVQI (3.01) and ABI, through external validation, has been shown to be a robust, objective, and ecologically validated tool for assessing overall voice quality and breathiness (Barsties & Maryn, 2016; Barsties v. Latoszek et al., 2021) which has been validated in several countries as cut-off scores vary between different languages due to variations in speech and its perceptual interpretation by raters (Maryn et al., 2014).

Next Steps: This study aims to evaluate the concurrent validity and diagnostic accuracy of the AVQI version 3.01 and ABI in American English speakers. We are looking for partners in this endeavor. If you are a clinician or researcher interested in being part of this process and contributing to the voice sample database, you can write to Adrian Castillo (casti208@msu.edu).

Updates
0 new
2
Recommendations
0 new
0
Followers
0 new
5
Reads
0 new
53

Project log

Adrián Castillo
added an update
This study aims to evaluate the concurrent validity and diagnostic accuracy of the AVQI version 3.01 in American English speakers. We are looking for partners in this endeavor.
If you are a clinician or researcher interested in being part of this process and contributing to the voice sample database, you can write to Adrian Castillo (casti208@msu.edu).
 
Adrián Castillo
added a project goal
Problem Statement and Impact: Instrumental voice assessment is a process that considers various methods of data collection, such as endoscopic assessment, aerodynamic assessment, self-assessment, auditory perceptual assessment, and objective acoustic voice analysis (Dejonckere et al., 2001; Patel et al., 2018). Measurement and assessment of the acoustic parameters of the voice have been one of the most commonly used approaches by voice clinicians as it is easily accessible and inexpensive to implement (Behlau & Murry, 2012; Mehta & Hillman, 2008). Currently, one of the dilemmas associated with acoustic voice analysis relates to the acoustic parameters of voice that one wishes to measure, which many of them underlie a particular type of voice sample, for example, the use of sustained vowel versus continuous speech tasks.
Background: Voice acoustic parameters come in many forms depending on the goal of the parameter, source of the acoustic signal, and the technique of analysis. Some parameters are more primary to voice production (e.g., fundamental frequency, loudness level), others are more related to voice quality (e.g., spectral slope, alpha ratio, CPP), others still are related to short term transient production effects (e.g., jitter, shimmer, relative fundamental frequency). Many of the voice acoustic parameters are obtained from the fundamental frequency (f0) of a sustained vowel, limiting the validity and reliability of acoustic measurements, so such measures should be interpreted with caution (Mehta & Hillman, 2008). Also, analysis of continuous speech samples has been shown to have more robust correlations with perceptual voice analysis (Lowell, 2012). However, it has also been postulated that speech samples alone may not accurately quantify what they are intended to measure due to the variability associated with speech characteristics, so their construct validity has also been questioned (Gerratt et al., 2016). Nevertheless, evidence has shown that software that allows concatenated analysis of audio samples, i.e., both continuous speech and a sustained vowel, provides reliable results concerning diagnostic accuracy and concurrent validity in overall voice quality when compared to auditory perceptual voice analysis (Barsties & De Bodt, 2015), the gold standard for assessing the severity of dysphonia (Maryn & Roy, 2012; Oates, 2009).
Overall Objective: The general aim of this study is to study the benefits and limitations of acoustic parameters when considering both steady production and running speech towards a more robust and valid assessment protocol that voice clinicians and voice scientists alike could use.
Current Goal: The specific goal of this study is further test the validity of a multidimensional analysis of voice called Acoustic Voice Quality Index (AVQI) Version 03.01 and the Acoustic Breathiness Index (ABI), which considers for its final result a multiple regression analysis of six concatenated acoustic variables, i.e., obtained from both a sample of sustained vowel and speech and thus generate a single score that predicts the severity of dysphonia (Batthyany et al., 2019; Maryn et al., 2010). The most recent version of the AVQI (3.01) and ABI, through external validation, has been shown to be a robust, objective, and ecologically validated tool for assessing overall voice quality and breathiness (Barsties & Maryn, 2016; Barsties v. Latoszek et al., 2021) which has been validated in several countries as cut-off scores vary between different languages due to variations in speech and its perceptual interpretation by raters (Maryn et al., 2014).
Next Steps: This study aims to evaluate the concurrent validity and diagnostic accuracy of the AVQI version 3.01 and ABI in American English speakers. We are looking for partners in this endeavor. If you are a clinician or researcher interested in being part of this process and contributing to the voice sample database, you can write to Adrian Castillo (casti208@msu.edu).