Joint Dictionary Learning-Based Non-Negative Matrix Factorization for Voice Conversion to Improve Speech Intelligibility After Oral Surgery.

Objective: This paper focuses on machine learning based voice conversion (VC) techniques for improving the speech intelligibility of surgical patients who have had parts of their articulators removed. Because of the removal of parts of the articulator, a patient's speech may be distorted and difficult to understand. To overcome this problem, VC methods can be applied to convert the distorted speech such that it is clear and more intelligible. To design an effective VC method, two key points must be considered: 1) the amount of training data may be limited (because speaking for a long time is usually difficult for postoperative patients); 2) rapid conversion is desirable (for better communication). Methods: We propose a novel joint dictionary learning based non-negative matrix factorization (JD-NMF) algorithm. Compared to conventional VC techniques, JD-NMF can perform VC efficiently and effectively with only a small amount of training data. Results: The experimental results demonstrate that the proposed JD-NMF method not only achieves notably higher short-time objective intelligibility (STOI) scores (a standardized objective intelligibility evaluation metric) than those obtained using the original unconverted speech but is also significantly more efficient and effective than a conventional exemplar-based NMF VC method. Conclusion: The proposed JD-NMF method may outperform the state-of-the-art exemplar-based NMF VC method in terms of STOI scores under the desired scenario. Significance: We confirmed the advantages of the proposed joint training criterion for the NMF-based VC. Moreover, we verified that the proposed JD-NMF can effectively improve the speech intelligibility scores of oral surgery patients.Objective: This paper focuses on machine learning based voice conversion (VC) techniques for improving the speech intelligibility of surgical patients who have had parts of their articulators removed. Because of the removal of parts of the articulator, a patient's speech may be distorted and difficult to understand. To overcome this problem, VC methods can be applied to convert the distorted speech such that it is clear and more intelligible. To design an effective VC method, two key points must be considered: 1) the amount of training data may be limited (because speaking for a long time is usually difficult for postoperative patients); 2) rapid conversion is desirable (for better communication). Methods: We propose a novel joint dictionary learning based non-negative matrix factorization (JD-NMF) algorithm. Compared to conventional VC techniques, JD-NMF can perform VC efficiently and effectively with only a small amount of training data. Results: The experimental results demonstrate that the proposed JD-NMF method not only achieves notably higher short-time objective intelligibility (STOI) scores (a standardized objective intelligibility evaluation metric) than those obtained using the original unconverted speech but is also significantly more efficient and effective than a conventional exemplar-based NMF VC method. Conclusion: The proposed JD-NMF method may outperform the state-of-the-art exemplar-based NMF VC method in terms of STOI scores under the desired scenario. Significance: We confirmed the advantages of the proposed joint training criterion for the NMF-based VC. Moreover, we verified that the proposed JD-NMF can effectively improve the speech intelligibility scores of oral surgery patients.

Changes in the nasalance of vowels within the first week following uvulopalatopharyngoplasty.

We conducted a retrospective study to evaluate the short-term effect of uvulopalatopharyngoplasty (UPPP) on changes in the nasalance of five vowels: /a/, /ε/, /i/, /ɔ/, and /u/. Our study group was made up of 20 patients-15 males and 5 females, aged 16 to 57 years (mean: 37.3 ± 11.5)-who had undergone UPPP as a treatment for mild to moderate obstructive sleep apnea. Nasometry was used to obtain nasalance scores in all patients on the morning of the operation (day 1) and subsequently on day 4 or 5 (mean: 4.5 ± 0.5). Preoperatively, nasalance scores were highest for /i/ (mean: 29.8 ± 12.6) and /a/ (mean: 24.1 ± 10.3). After the operation, nasalance scores for all five studied vowels increased; they were highest for /i/ (mean: 40.7 ± +17.8) and /ε/ (mean: 30.0 ± 10.8). The increases in the nasalance of /i/, /ε/, and /u/ were statistically significant (p < 0.05). Our findings indicate that UPPP has a significant impact on nasalance immediately after surgery.

The natural course of congenital nasal pyriform aperture stenosis.

OBJECTIVES/HYPOTHESIS: Patients with congenital nasal pyriform aperture stenosis (CNPAS) may become less symptomatic with age. Therefore, we aimed to develop a growth curve of the pyriform aperture so that a more comprehensive plan can be designed for CNPAS patients who show little response to conservative treatment. STUDY DESIGN: A single-institution study, retrospective review of CNPAS patients during the period November 1997 to December 2014. METHODS: We measured the distances between the bilateral nasal processes of the maxilla (interprocess distance [IPD]) on three-dimensional computed tomography images and then divided the patients into five different age groups. A growth curve of the pyriform aperture was then constructed based on the distance-age relationship. RESULTS: Fifty-four IPD measurements were included. The mean IPD was 3.57 mm in neonates < 1 month old, 4.08 mm in infants aged 1 to 3 months, 5.19 mm in the 4-month to 11-month age group, 6.61 mm in the 12-month to 36-month age group, and 9.20 mm in children > 36 months of age. We found that the cubic curve was the most appropriate growth curve, and that growth tended to be slower from 3.5 years to 6 years of age. CONCLUSIONS: The growth curve of the pyriform aperture in children with CNPAS developed in this study can aid in treatment planning and predict clinical outcome of CNPAS patients. Although CNPAS patients may become less symptomatic with age, when the observed IPD falls progressively farther from the curve, more aggressive intervention should be considered, such as changing the management strategy from observation to conservative treatment or from conservative treatment to surgery. LEVEL OF EVIDENCE: 4 Laryngoscope, 126:2399-2402, 2016.

Subjective neck pain or foreign body sensation and the true location of foreign bodies in the pharynx.

CONCLUSION: The location of the foreign body did not correspond well to the location of pain reported by patients. When patients present with foreign bodies in the pharynx, in addition to recording the location of pain and foreign body sensation, clinicians should perform a comprehensive and thorough oropharyngeal examination to avoid misdiagnosis. OBJECTIVES: Physicians are often guided by patient-indicated locations of pharyngeal foreign bodies. In this study, we aimed to determine the correlation between the location of the subjective neck pain or foreign body sensation and the true location of the foreign body. METHODS: We prospectively studied 90 patients who had pharyngeal foreign bodies removed at MacKay Memorial Hospital. We divided the head and neck into 10 zones according to the superficial anatomy. Subjective location, examination findings, and actual foreign body location were recorded and compared. RESULTS: The overall subjective and true locations of the foreign body were poorly correlated (kappa 0.27, p = 0.003). The positive predictive value (PPV) for the midline neck was 68%, which was higher than that on either lateral side of the neck. PPV above cricoid cartilage level was 66%.

A simple algorithm for treating horizontal benign paroxysmal positional vertigo.

OBJECTIVE: Horizontal benign paroxysmal positional vertigo (H-BPPV) is more difficult to successfully treat than posterior benign paroxysmal positional vertigo (P-BPPV) because of the diverse mechanisms required. We developed a simple, rapid, and effective treatment algorithm for treating all subtypes of H-BPPV in an ear, nose, and throat (ENT) outpatient department. MATERIALS AND METHODS: Four hundred ninety patients with BPPV receiving outpatient treatment at Mackay Memorial Hospital were investigated. Among the 490 patients, 86 (17.6%; 86/490) were diagnosed as having H-BPPV variants using the McClure-Pagnini test. Fifty-four patients were female, and 32 were male; they ranged in age from 18 to 92 years (mean age, 56.2 yr). RESULTS: Among the 86 H-BPPV patients, 74.4% (64/86) were hypothesized to have canalithiasis, 20.9% (18/86) were hypothesized to have cupulolithiasis-utricle type (Cup-U), and 4.7% (4/86) were hypothesized to have the cupulolithiasis-cupula type (Cup-C). The primary treatment maneuver was the forced prolonged position (FPP). For 3 patients exhibiting refractory symptoms, we introduced the Gufoni maneuver. The total average success rate of treatment was 96%. CONCLUSION: We concluded that for H-BPPV patients with initial geotropic nystagmus, the FPP alone yielded an excellent treatment-control rate, and the barbecue-rotation maneuver was unnecessary. However, observing the nystagmus transformation of apogeotropic patients was necessary before administering treatment. For cupulolithiasis patients with the apogeotropic variant who did not respond to FPP treatment alone, we determined that the Gufoni maneuver was necessary as well.

Acoustic characteristics of different target vowels during the laryngeal telescopy.

OBJECTIVE: The aim of this study was to investigate the acoustic characteristics of target vowels phonated in normal voice persons while performing laryngeal telescopy. The acoustic characteristics are compared to show the extent of possible difference to speculate their impact on phonation function. METHODS: Thirty-four male subjects aged 20-39 years with normal voice were included in this study. The target vowels were /i/ and /ɛ/. Recording of voice samples was done under natural phonation and during laryngeal telescopy. The acoustic analysis included the parameters of fundamental frequency, jitter, shimmer and noise-to-harmonic ratio. RESULTS: The sound of a target vowel /ɛ/ was perceived identical in more than 90% of the subjects by the examiner and speech language pathologist during the telescopy. Both /i/ and /ɛ/ sounds showed significant difference when compared with the results under natural phonation. There was no significant difference between /i/ and /ɛ/ during the telescopy. CONCLUSION: The present study showed that change in target vowels during laryngeal telescopy makes no significant difference in the acoustic characteristics. The results may lead to the speculation that the phonation mechanism was not affected significantly by different vowels during the telescopy. This study may suggest that in the principle of comfortable phonation, introduction of the target vowels /i/ and /ɛ/ is practical.

The effectiveness of the promotion of newborn hearing screening in Taiwan.

OBJECTIVE: Hearing is a critical ability for the development of a child's speech and language. Many studies in different countries have shown the universal newborn hearing screening and early intervention has greatly reduced the negative impact caused by congenital hearing loss. The first universal newborn hearing screening program in Taiwan took place in MacKay Memorial Hospital in 1998 and was subsequently endorsed by the government. The incidence of bilateral congenital hearing impairment in Taiwan is approximately 2.6 per 1000 live birth. The aim of this paper is to analyze the age of diagnosis, hearing aid fitting, and intervention of congenitally hearing impaired children with and without hearing screening after public awareness and government endorsement of newborn hearing screening. MATERIALS AND METHODS: There were 263 hearing impaired children participated in this study, receiving their auditory habilitation therapy at Children's Hearing Foundation from 2006 to 2010. 114 of those children went through newborn hearing screening and 149 without it. The age of diagnosis, hearing aid fitting, and auditory intervention were compared between these two groups. The age of diagnosis and intervention of congenitally hearing impaired children among different years were analyzed too. RESULTS: The average age of diagnosis was 8.7 months, the age of hearing aid fitting was 12.4 months and age of auditory intervention was 18.8 months for the group of hearing impaired children with newborn hearing screening. For hearing impaired children without newborn screening, their average age of diagnosis was 27.5 months; age of hearing aid fitting was 31.3 months and age of auditory intervention was 40.5 months. There were significant differences in the age of diagnosis, hearing aid fitting and auditory intervention between congenitally hearing impaired children with and without hearing screening. CONCLUSIONS: This research indicates that newborn hearing screening facilitates early identification, diagnosis and intervention of congenitally hearing impaired children in Taiwan. The age of identification, diagnosis and intervention of congenital hearing impaired children has also been reduced gradually over the years after government endorsement of newborn hearing screening in Taiwan.

The outcomes of infantile vallecular cyst post CO2 laser treatment.

OBJECTIVES: Vallecular cyst is not a common disease of neonate and infant. However, it may cause severe airway obstruction and even death. Its clinical symptoms are similar to laryngomalacia, including stridor, suprasternal retraction, substernal retraction, feeding difficulties, vomiting, failure to thrive, feeding choking and desaturation. This study is aimed to evaluate the surgical outcomes of infantile vallecular cyst post CO2 laser treatment and to explore the appropriate time point of surgery for infantile vallecular cyst. METHODS: In a retrospective review, thirty three patients diagnosed as vellecular cyst were enrolled in this study. All the patients received awake fiberoptic videobronchoscopic examination in order to prove the diagnosis. Pre-operative and post-operative eight symptom items were both recorded for comparison of the surgical outcomes. The age of diagnosis, gender, operation, body weight at surgery, co-morbidities, dates of postoperative endotracheal intubation, ICU stays and admission days were all recorded for analysis. RESULTS: Total 33 cases of vallecular cyst were diagnosed by fiberoptic videobrochoscopy at our department. Most infants were diagnosed at 2-3 months-old. Stridor was the most common pre-operative symptom (100%). Additionally, feeding choking was the most common post-operative one (29.41%). Two patients (6.06%) with newly onset postoperative feeding choking recovered spontaneously within 2 weeks. Failure to thrive got worse as age increased before the operation. Eighty-eight percent of patients had good or excellent improvement of symptoms after surgery. There was neither recurrence nor surgical complication in our study. CONCLUSION: Vallecular cyst often combined with laryngomalacia. Different from laryngomalacia, it had excellent outcomes after CO2 laser treatment. Failure to thrive got worse if delay diagnosis. Therefore, if definite diagnosis is made, early laser excision of cyst is a good method and the surgical outcomes are excellent.

A fabricated forearm free flap with accompanying phonation tube for simultaneous reconstruction of a pharyngolaryngeal circumferential defect and voice loss: new surgical modification with functional phonation outcome.

BACKGROUND: Total pharyngolaryngectomy is potentially ablative surgery, resulting in compromise of some most basic functions of life, including speech and swallowing. Tracheoesophageal puncture is the gold standard for voice restoration. But it still has prosthesis-related problems. OBJECTIVES/HYPOTHESIS: We designed a uniquely customized radial forearm free flap (RFFF), which also incorporated a region for phonation tube (PT) creation, for the dual purpose of circumferential laryngopharyngeal defect reconstruction and voice production. METHODS: From August 2005 to September 2010, there were 18 male patients with late-stage hypopharyngeal cancer (HPC) or laryngeal cancer (LC) who received one-stage reconstruction with the fabricated RFFF-accompanying PT after total pharyngolaryngectomy. We recorded the phonation outcome of phonation efficacy (PE) and maximal phonation time (MPT) postoperatively within 1 month and at least 1 year after surgery. RESULTS: Nine patients suffered from HPC and the others suffered from LC. Twelve patients received concurrent chemoradiotherapy after surgery. The follow-up time was 12 to 56 months (mean 28.7 months). There was no significant variance in the PE (79.72%, SD=21.93% vs. 62.50%, SD=39.60%, respectively; p = 0.115) and MPT (2.58 seconds, SD=1.80 vs. 2.97 seconds, SD=3.96, respectively; p = 0.878) between the first and last follow-up points, even when the patients were grouped by radiotherapy status after surgery or by disease group. CONCLUSIONS: The phonation outcome in our experience was satisfactory and it tolerated postoperative radiotherapy during at least the 12-month follow-up period.

Voice and speech outcomes with radial forearm free flap-accompanied phonation tube after total pharyngolaryngectomy of hypopharyngeal cancer.

CONCLUSIONS: Radial forearm free flap (RFFF)-accompanied phonation tube (PT) for voice and speech restoration after pharyngolaryngectomy is promising, especially in phonation efficacy and intelligibility. It offers not only another safe surgical option but also a satisfactory result for such patients with advanced hypopharyngeal cancer. OBJECTIVES: We use RFFF with RFFF-accompanied PT for one-stage reconstruction both for tissue defect and voice reconstruction in patients undergoing total pharyngolaryngectomy. METHODS: Eight male patients with advanced hypopharyngeal cancer underwent total pharyngolaryngectomy. Voice restoration was done with RFFF-accompanied PT. Phonation outcomes and speech outcomes of the patients were evaluated and scored. RESULTS: The mean follow-up time was 13.7 months. All free flaps were successful without perioperative mortality. All the patients were able to produce sound. Phonation efficacy ranged from 70% to >90% postoperatively and 40% to >90% at the last follow-up. The speech intelligibility was graded as moderately good.

Other impairment associated with developmental language delay in preschool-aged children.

The aim of this study was to investigate developmental impairment in several domains that might be associated with developmental language delay. The records of 56 preschool children with developmental language delay and 31 nonimpaired children were reviewed. Children with language delay were more likely than those in the nonimpaired group to have cognitive developmental delay (Mental Development Index < 70) (P < .001) and gross and fine motor delay (gross: 28 [50%] versus 5 [16%], P = .002; fine: 34 [62%] versus 11 [35%], P = .02). Children with language delay were significantly more likely to have impairment than were nonimpaired children in gross motor, fine motor, comprehension-conceptual and personal-social (P = .01, P = .02, P = .01, P = .02, respectively) functional domains. Our findings indicate that preschool children with language delay have wide-ranging difficulties in development and function.

Effect of identification and intervention age on language development for Mandarin-speaking deaf children with high family involvement.

OBJECTIVE: The purpose of this study is to assess the language ability between early-intervention and later-intervention Mandarin-speaking deaf children, who have normal cognition and high family involvement. MATERIALS AND METHODS: There are 29 subjects enrolled. 11 born deaf children received early intervention (7 HA and 4 CI) before 6 months old as study group. Another 18 born deaf children received later intervention (11 HA and 7 CI) between 7 and 35 months old as reference group. They were all regarded as with normal cognition and high family involvement. Their mean assessment age was 50 months old in early group and 51 months old in later group. We used several tools to test their perceptive vocabulary size, to evaluate perceptive language syntax and to compare perceptive and expressive language scores. RESULTS: Our study revealed there are significant difference between these two groups in the ability of vocabulary size, perceptive language syntax and perceptive language scores. The results showed there is no significant difference between these two groups in their expressive language scores, although their achievement score is higher in the early group. CONCLUSIONS: It clearly showed the ability of perceptive language in early-intervention deaf children was better than that of later-intervention. The ability of their expressive language showed no difference between them.

Cholesterol granuloma of the external ear canal: A rare presentation.

Almost all aural cholesterol granulomas develop in the mucosa of the middle ear. We describe the case of a 14-year-old girl who presented with an asymptomatic, nontender, dark-blue, cystic lesion in the posteroinferior portion of the left ear canal. The mass was excised via a postauricular approach. Postoperatively, the mass was identified as a cholesterol granuloma on pathologic examination. To the best of our knowledge, this is the first reported case of a human cholesterol granuloma limited to the external ear canal.