An official website of the United States government
In realistic speech enhancement settings for end-user devices, we often encounter only a few speakers and noise types that tend to reoccur in the specific acoustic environment. We propose a novel personalized speech enhancement method to adapt a compact denoising model to the test-time specificity. Our goal in this test-time adaptation is to utilize no clean speech target of the test speaker, thus fulfilling the requirement for zero-shot learning. To complement the lack of clean speech, we employ the knowledge distillation framework: we distill the more advanced denoising results from an overly large teacher model, and use them as the pseudo target to train the small student model. This zero-shot learning procedure circumvents the process of collecting users' clean speech, a process that users are reluctant to comply due to privacy concerns and technical difficulty of recording clean voice. Experiments on various test-time conditions show that the proposed personalization method can significantly improve the compact models' performance during the test time. Furthermore, since the personalized models outperform larger non-personalized baseline models, we claim that personalization achieves model compression with no loss of denoising performance. As expected, the student models underperform the state-of-the-art teacher models.
more »
« less
Listening to Sounds of Silence for Speech Denoising
We introduce a deep learning model for speech denoising, a long-standing challenge in audio analysis arising in numerous applications. Our approach is based on a key observation about human speech: there is often a short pause between each sentence or word. In a recorded speech signal, those pauses introduce a series of time periods during which only noise is present. We leverage these incidental silent intervals to learn a model for automatic speech denoising given only mono-channel audio. Detected silent intervals over time expose not just pure noise but its time-varying features, allowing the model to learn noise dynamics and suppress it from the speech signal. Experiments on multiple datasets confirm the pivotal role of silent interval detection for speech denoising, and our method outperforms several state-of-the-art denoising methods, including those that accept only audio input (like ours) and those that denoise based on audiovisual input (and hence require more information). We also show that our method enjoys excellent generalization properties, such as denoising spoken languages not seen during training.
more »
« less
- PAR ID:
- 10414132
- Date Published:
- Journal Name:
- Advances in neural information processing systems
- ISSN:
- 1049-5258
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Recently, Deep Image Prior (DIP) has emerged as an effective unsupervised one-shot learner, delivering competitive results across various image recovery problems. This method only requires the noisy measurements and a forward operator, relying solely on deep networks initialized with random noise to learn and restore the structure of the data. However, DIP is notorious for its vulnerability to overfitting due to the overparameterization of the network. Building upon insights into the impact of the DIP input and drawing inspiration from the gradual denoising process in cutting-edge diffusion models, we introduce Autoencoding Sequential DIP (aSeqDIP) for image reconstruction. This method progressively denoises and reconstructs the image through a sequential optimization of network weights. This is achieved using an input-adaptive DIP objective, combined with an autoencoding regularization term. Compared to diffusion models, our method does not require training data and outperforms other DIP-based methods in mitigating noise overfitting while maintaining a similar number of parameter updates as Vanilla DIP. Through extensive experiments, we validate the effectiveness of our method in various image reconstruction tasks, such as MRI and CT reconstruction, as well as in image restoration tasks like image denoising, inpainting, and non-linear deblurring.more » « less
-
Silent speech is unaffected by ambient noise, increases accessibility, and enhances privacy and security. Yet current silent speech recognizers operate in a phrase-in/phrase-out manner, thus are slow, error prone, and impractical for mobile devices. We present MELDER, a Mobile Lip Reader that operates in real-time by splitting the input video into smaller temporal segments to process them individually. An experiment revealed that this substantially improves computation time, making it suitable for mobile devices. We further optimize the model for everyday use by exploiting the knowledge from a high-resource vocabulary using a transfer learning model. We then compare MELDER in both stationary and mobile settings with two state-of-the-art silent speech recognizers, where MELDER demonstrated superior overall performance. Finally, we compare two visual feedback methods of MELDER with the visual feedback method of Google Assistant. The outcomes shed light on how these proposed feedback methods influence users' perceptions of the model's performance.more » « less
-
Human speech perception is generally optimal in quiet environments, however it becomes more difficult and error prone in the presence of noise, such as other humans speaking nearby or ambient noise. In such situations, human speech perception is improved by speech reading , i.e., watching the movements of a speaker's mouth and face, either consciously as done by people with hearing loss or subconsciously by other humans. While previous work focused largely on speech perception of two-dimensional videos of faces, there is a gap in the research field focusing on facial features as seen in head-mounted displays, including the impacts of display resolution, and the effectiveness of visually enhancing a virtual human face on speech perception in the presence of noise. In this paper, we present a comparative user study ( $N=21$ ) in which we investigated an audio-only condition compared to two levels of head-mounted display resolution ( $$1832\times 1920$$ or $$916\times 960$$ pixels per eye) and two levels of the native or visually enhanced appearance of a virtual human, the latter consisting of an up-scaled facial representation and simulated lipstick (lip coloring) added to increase contrast. To understand effects on speech perception in noise, we measured participants' speech reception thresholds (SRTs) for each audio-visual stimulus condition. These thresholds indicate the decibel levels of the speech signal that are necessary for a listener to receive the speech correctly 50% of the time. First, we show that the display resolution significantly affected participants' ability to perceive the speech signal in noise, which has practical implications for the field, especially in social virtual environments. Second, we show that our visual enhancement method was able to compensate for limited display resolution and was generally preferred by participants. Specifically, our participants indicated that they benefited from the head scaling more than the added facial contrast from the simulated lipstick. We discuss relationships, implications, and guidelines for applications that aim to leverage such enhancements.more » « less
-
Human speech perception is generally optimal in quiet environments, however it becomes more difficult and error prone in the presence of noise, such as other humans speaking nearby or ambient noise. In such situations, human speech perception is improved by speech reading, i.e., watching the movements of a speaker’s mouth and face, either consciously as done by people with hearing loss or subconsciously by other humans. While previous work focused largely on speech perception of two-dimensional videos of faces, there is a gap in the research field focusing on facial features as seen in head-mounted displays, including the impacts of display resolution, and the effectiveness of visually enhancing a virtual human face on speech perception in the presence of noise. In this paper, we present a comparative user study (N = 21) in which we investigated an audio-only condition compared to two levels of head-mounted display resolution (1832×1920 or 916×960 pixels per eye) and two levels of the native or visually enhanced appearance of a virtual human, the latter consisting of an up-scaled facial representation and simulated lipstick (lip coloring) added to increase contrast. To understand effects on speech perception in noise, we measured participants’ speech reception thresholds (SRTs) for each audio-visual stimulus condition. These thresholds indicate the decibel levels of the speech signal that are necessary for a listener to receive the speech correctly 50% of the time. First, we show that the display resolution significantly affected participants’ ability to perceive the speech signal in noise, which has practical implications for the field, especially in social virtual environments. Second, we show that our visual enhancement method was able to compensate for limited display resolution and was generally preferred by participants. Specifically, our participants indicated that they benefited from the head scaling more than the added facial contrast from the simulated lipstick. We discuss relationships, implications, and guidelines for applications that aim to leverage such enhancements.more » « less
- Free Publicly Accessible Full Text
-
Accepted Manuscript1.0
- Conference Paper:
- The DOI is not currently available.
