通过解散细粒度风格建模来改进零拍摄风格转移文本到语音的方法
Eray Eren1, Qingju Liu2, Abeer Alwan1
1Department of Electrical and Computer Engineering, University of California, Los Angeles, California 90095, USA.
JASA express letters
|March 10, 2026
概括
本研究介绍了一种新的零射击风格转移语音合成方法,使用一个小的GenerSpeech模型. 该方法增强了对未见的说话风格的风格适应,优于现有的方法.
科学领域:
- 语音合成 语音合成
- 人工智能的人工智能
- 机器学习 机器学习
背景情况:
- 零射击风格转移语音合成旨在适应看不见的说话风格.
- 当前最先进的方法通常需要大型模型或广泛的数据来进行概括.
- 通过较小的模型实现有效的概括仍然是一个重大挑战.
研究的目的:
- 提出一种新的零射击风格转移语音合成方法,利用一个紧的GenerSpeech骨干.
- 提高语音合成模型的适应能力,以适应资源有限的新型说话风格.
- 解开扬声器,风格和内容嵌入,以提高合成质量.
主要方法:
- 一个细粒度的风格编码器与GenerSpeech骨干集成.
- 相互信息最小化损失被用来解开扬声器,全球/细粒度风格和内容嵌入的纠.
- 引入了最大平均差异引导周期的一致性损失,以进一步将风格与扬声器分开,并增加风格嵌入多样性.
主要成果:
- 与已建立的零射击风格转移基线 (GenerSpeech,YourTTS,VALL-E-X) 相比,拟议的方法显示出更高的性能.
- 实现了31%的相对平均风格偏好改进.
- 在VCTK数据集上获得了3.64的prosody相似度的平均意见得分.
结论:
- 开发的方法有效地实现了零射击的风格转移语音合成与一个小模型.
- 拟议的脱损失显著改善了风格适应和合成质量.
- 这项工作提升了小型模型对各种说话风格的概括能力.
相关概念视频
02:07Improving Translational Accuracy
3.7K
3.7K
02:07Improving Translational Accuracy
15.3K
Base complementarity between the three base pairs of mRNA codon and the tRNA anticodon is not a failsafe mechanism. Inaccuracies can range from a single mismatch to no correct base pairing at all. The free energy difference between the correct and nearly correct base pairs can be as small as 3 kcal/ mol. With complementarity being the only proofreading step, the estimated error frequency would be one wrong amino acid in every 100 amino acids incorporated. However, error frequencies observed in...
15.3K
01:14Impression Management Techniques IV: Altercasting
207
Altercasting is a strategic communication technique in which an individual imposes a specific identity or social role onto another person to influence their behavior and shape the interaction. By presuming a role—such as “responsible leader” or “patient person”—altercasting encourages the target to conform to that identity, often aligning their behavior with the expectations associated with the role. The power of this tactic lies in its subtlety; once a role...
207
02:16Stereotype Content Model
15.6K
The Stereotype Content Model (SCM) was first proposed by Susan Fiske and her colleagues (Fiske, Cuddy, Glick & Xu, 2002; see also Fiske, 2012 and Fiske, 2017). The SCM specifies that when someone encounters a new group, they will stereotype them based on two metrics: warmth—or that group’s perceived intent, and how likely they are to provide help or inflict harm—and competence—or their ability to carry out that objective. Depending on the warmth-competence...
15.6K
01:12Modeling and Similitude
699
Scaled modeling is a fundamental technique in engineering, enabling the study of large and complex systems by creating smaller, manageable replicas that recreate critical characteristics of the original. In hydrology and civil infrastructure, for example, scaled models of dams help analyze water flow, turbulence, and pressure. This method allows for accurate predictions of real-world behavior within a controlled environment, significantly reducing the cost and time involved in full-scale...
699
01:22Language Development
1.0K
Children master language quickly and with relative ease, supported by both biological predisposition and reinforcement. B. F. Skinner (1957) proposed that language is learned through reinforcement, while Noam Chomsky (1965) argued that language acquisition mechanisms are biologically determined.
The critical period for language acquisition suggests that the ability to acquire language is at its peak early in life. As people age, this proficiency decreases. Language development begins very...
The critical period for language acquisition suggests that the ability to acquire language is at its peak early in life. As people age, this proficiency decreases. Language development begins very...
1.0K