量化自动语音识别系统在d/聋人和听力障碍者的语音表现
Robin Zhao1, Anna S G Choi2, Allison Koenecke2
1Sean Parker Institute for the Voice, Weill Cornell Medical College, New York, New York, U.S.A.
The Laryngoscope
|August 19, 2024
概括
商业自动语音识别 (ASR) 系统在听障和听力障碍 (d/Dhh) 个体的语音上遇到了困难. 对于那些语音理解能力较低,语言前听力损失和手语作为主要沟通方式的人来说,表现明显更差.
科学领域:
- 语音处理和声学学
- 人与计算机的互动.
- 可访问性技术可访问性技术
背景情况:
- 商业自动语音识别 (ASR) 系统被广泛使用,但它们对各种语音模式的性能,特别是来自聋人和听力障碍者 (d/Dhh) 个人的表现仍然被低估.
- 现有的ASR技术可能无法充分代表d/Dhh人群的语音特征,可能会限制可访问性.
研究的目的:
- 评估领先的商业自动语音识别 (ASR) 系统在处理d/Dhh个人的语音时的性能.
- 为了确定d/Dhh群体中影响ASR准确性的特定因素.
主要方法:
- 来自d/Dhh和正常听力 (NH) 参与者的850个音频文件的语料库在四个主要的ASR API中进行了测试:亚马逊网络服务,微软Azure,Google Chirp和OpenAI Whisper.
- 文字错误率 (WER) 量化了24d/Dhh和9名NH参与者,根据语音可理解性分类 (SIC),听力损失 (HL) 发病和主要通信模式进行了分组分析.
主要成果:
- 与NH组 (5.0%) 相比,d/Dhh组的平均WER是d/Dhh组的10倍 (52.6%).
- 在较低的SIC水平 (85.9%为"低",46.6%为"中等") 和"高"SIC (9.5%) 之间,ASR表现显著下降.
- 预语言HL (80.5%WER) 和主要使用手语 (70.2%WER) 的说话者表现出比后语言HL或仅口语沟通的错误率高得多.
结论:
- 商业ASR系统对d/Dhh个体表现明显不佳,特别是那些语音理解能力较低,语言前听力损失或手语作为主要沟通方式的人.
- 这一绩效差距凸显了与主要技术公司的可访问性要求存在差异.
- 需要开发ASR系统并使用各种d/Dhh语音数据进行伦理训练,以确保公平的可访问性.
更多相关视频
11:39Assessment of Audio-Tactile Sensory Substitution Training in Participants with Profound Deafness Using the Event-Related Potential Technique
Published on: September 7, 2022
2.1K
06:04Systematic Hearing Performance Evaluation Process for Adolescents with Cochlear Implantation at Early Ages
Published on: March 24, 2023
359
相关概念视频
01:31Hearing
48.0K
When we hear a sound, our nervous system is detecting sound waves—pressure waves of mechanical energy traveling through a medium. The frequency of the wave is perceived as pitch, while the amplitude is perceived as loudness.
48.0K
01:17Auditory Perception
1.5K
The auditory system is essential for sound perception, utilizing various critical structures. When sound waves enter the outer ear, they travel through the ear canal and cause the eardrum to vibrate. These vibrations are then transmitted to the middle ear, where three tiny bones – the malleus, incus, and stapes – amplify the sound. This amplification is crucial, as it ensures that the sound vibrations are strong enough to be conveyed to the inner ear. These vibrations then reach the...
1.5K
01:25Learning Disabilities
697
Learning disabilities are cognitive disorders caused by neurological impairments that affect cognitive functions like language and reading, without indicating overall intellectual or developmental challenges. These disabilities differ from global intellectual or developmental disabilities as they are limited to distinct cognitive functions. Common learning disabilities include dysgraphia, dyslexia, and dyscalculia, each of which impacts unique aspects of learning.
Dyslexia
Dyslexia is a...
Dyslexia
Dyslexia is a...
697