基于跨域传播的语音增强,用于非常杂的语音
Heming Wang1, DeLiang Wang1,2
1Department of Computer Science and Engineering, The Ohio State University, USA.
概括
这项研究引入了一种用于语音增强的新型深度学习方法,通过整合基于扩散的学习,在低信号噪声比 (SNR) 条件下显著提高性能. 该方法增强了对非静止噪声的稳定性,在极端杂的环境中性能优于现有的技术.
科学领域:
- 人工智能的人工智能
- 信号处理 信号处理
- 机器学习 机器学习
背景情况:
- 深度学习已经推进了语音增强,但仍在与低信号对噪声比 (SNR) 和非静止噪声作斗争.
- 强大的语音增强对于在不利的声学环境中的应用至关重要.
研究的目的:
- 开发一个更强大的深度学习模型来增强语音,特别有效在极端杂的条件下.
- 通过结合基于扩散的生成学习,改善低SNR场景中的语音恢复.
主要方法:
- 开发了一个基于频域扩散的生成模块.
- 该模块使用来自时间域监督增强模块的增强信号作为辅助输入.
- 该模型被训练来恢复清洁的语音谱图.
主要成果:
- 拟议的模型表明,与强大的基线相比,语音增强性能优越.
- 在极端杂的环境中观察到显著改善,SNR水平为-5dB和-10dB.
- 在TIMIT数据集上进行了实验.
结论:
- 整合基于扩散的学习增强了深度学习语音增强模型的稳定性.
- 拟议的方法有效地解决了低SNR和非静止噪声带来的挑战.
- 这种方法为未来关于强大的语音处理的研究提供了有希望的方向.
更多相关视频
06:04Systematic Hearing Performance Evaluation Process for Adolescents with Cochlear Implantation at Early Ages
Published on: March 24, 2023
311
09:09Foreign Accent and Forensic Speaker Identification in Voice Lineups: The Influence of Acoustic Features Based on Prosody
Published on: September 27, 2024
378
相关概念视频
Downsampling
110
When considering a sampled sequence with zero values between sampling instants, one can replace it by taking every N-th value of the sequence. At these integer multiples of N, the original and sampled sequences coincide. This process, known as decimation, involves extracting every N-th sample from a sequence, thereby creating a more efficient sequence.
The Fourier transform of the decimated sequence reveals a combination of scaled and shifted versions of the original spectrum. This...
The Fourier transform of the decimated sequence reveals a combination of scaled and shifted versions of the original spectrum. This...
110
Perceiving Loudness, Pitch, and Location
163
The human brain perceives pitch through two primary mechanisms reflected in place theory and frequency theory. Each mechanism describes how sound waves are interpreted as specific pitches by the brain, offering insights into the intricate processes of auditory perception.
Place theory, or place coding, suggests that different pitches are heard because various sound waves activate specific locations along the cochlea's basilar membrane. The brain determines the pitch of a sound by...
Place theory, or place coding, suggests that different pitches are heard because various sound waves activate specific locations along the cochlea's basilar membrane. The brain determines the pitch of a sound by...
163
Difference from Background: Limit of Detection
4.7K
The limit of detection (LOD) is the smallest amount of analyte that can be distinguished from the background noise. The LOD value corresponds to the concentration at which the analyte signal is three times larger than the standard deviation of the blank signal. Below this value, the analyte signal cannot be differentiated from the background noise. It is calculated by dividing the calibration slope by 3 times the standard deviation of the blank signals.
The LOD indicates the presence or absence...
The LOD indicates the presence or absence...
4.7K
Linear Approximation in Frequency Domain
79
Linear systems are characterized by two main properties: superposition and homogeneity. Superposition allows the response to multiple inputs to be the sum of the responses to each individual input. Homogeneity ensures that scaling an input by a scalar results in the response being scaled by the same scalar.
In contrast, nonlinear systems do not inherently possess these properties. However, for small deviations around an operating point, a nonlinear system can often be approximated as linear....
In contrast, nonlinear systems do not inherently possess these properties. However, for small deviations around an operating point, a nonlinear system can often be approximated as linear....
79
Reconstruction of Signal using Interpolation
149
Signal processing techniques are essential for accurately converting continuous signals to digital formats and vice versa. When a continuous signal is sampled with a period T, the resulting sampled signal exhibits replicas of the original spectrum in the frequency domain, spaced at intervals equal to the sampling frequency. To handle this sampled signal, a zero-order hold method can be applied, which creates a piecewise constant signal by retaining each sample's value until the next...
149
Unsoundness of Aggregate due to Volume Change
87
Unsoundness in aggregates due to volume changes is primarily caused by the physical alterations aggregates undergo, such as freezing and thawing, thermal changes, and wetting and drying. Unsound aggregates, when subjected to these changes, result in volume change upon disintegration. This, in turn, contributes to the deterioration of concrete, including scaling, pop-outs, and cracking. Particular types of aggregates, such as porous flints, cherts, and those containing clay minerals, are...
87
