Publications
D. Luong, K. Drossos, M. Heikkinen, and T. Virtanen, "Automatic Contextual Audio Denoising," in Proceedings of 34th European Signal Conference (EUSIPCO), Bruges, Belgium, 2026
Audio context determines which sound components and sources are relevant and which can be perceived as irrelevant (noise) by listeners. For example, traffic noise is informative in urban surveillance but noise for a phone call at the same location. Most current audio denoising systems apply fixed target-noise definitions, often removing useful components in one context while failing to suppress irrelevant components. To address this, we introduce the concept automatic contextual audio denoising (ACAD) which defines target and noise based on the inferred context. In this work, we restrict context to be associated with an acoustic scene class. We label sound events outside the event distribution of a scene class (noise) as out-of-context (OC) and events typical for that scene as in-context (IC). We implement a deep learning method that automatically infers the context of the audio signal and removes OC components, and benchmark it against variants: without context inference, with oracle context, and with separately provided uninformative context. On paired clean/noisy data across diverse contexts, where OC components in one context may be IC in another, our proposed method outperforms other approaches across standard objective metrics, indicating that the model can infer context and context-dependent processing can enhance denoising.
This paper introduces automatic contextual audio denoising (ACAD), where what counts as “noise” depends on the situation rather than being fixed in advance. For example, traffic sounds may be unwanted during a phone call but useful in an urban monitoring application. The proposed system first infers the acoustic context of the recording, such as whether it comes from a street, restaurant, kitchen, or subway, and then uses that context to decide which sounds are relevant and which are out of context and should be removed. To study this problem, the paper also constructs a dataset in which the same type of sound can be considered noise in one context but useful in another, forcing the model to make genuinely context-dependent decisions. The method combines a context-recognition network with a neural denoiser whose processing is guided by the learned context. Experiments show that using inferred context improves denoising compared with a conventional context-independent system, and the best version improves SI-SDR by almost 2 dB over the standard baseline. Interestingly, the learned context performs even better than simply providing the correct scene label, suggesting that the model captures richer information about the acoustic situation than a predefined class alone. The results support the broader idea that future audio-processing systems can benefit from understanding where and under what circumstances a sound occurs before deciding what should be preserved or removed.
K. Drosos, M. O. Heikkinen, J. T. Vilkamo, P. Tsiaflakis, “Model for speech enhancement,” U.S. Patent US20260065922A1, filed Aug. 15, 2025; published Mar. 05, 2026
Examples of the disclosure relate to a model that can be used for speech enhancement. The model comprises an encoder part comprising a sequence of encoding layers and caused to receive input data. The input data is based on a current frame of a noisy speech signal and one or more past frames of the noisy speech signal. The sequence of encoding layers is caused to process the input data so that output data of the encoder part comprises a reduced number of the multiple frequency positions and a single temporal position. The model also comprises a decoder part comprising a sequence of decoding layers caused to receive data from a prior decoding layer. The output data of the decoder part comprises multiple frequency positions and a single temporal position. The output data of the decoder part is for post processing to provide an output signal for speech enhancement.
This patent presents a computationally efficient neural-network architecture for speech enhancement, targeting real-time use on devices with limited processing and memory resources. The approach uses a causal U-Net that processes the current noisy speech frame together with information from previous frames, allowing it to exploit both short- and longer-term speech patterns without processing a large time-frequency region throughout the entire network. During encoding, the representation is progressively compressed until only a single temporal position remains, while the decoder reconstructs the required frequency resolution using skip connections from corresponding encoder layers. A post-processing stage can then use the resulting features to estimate an enhancement mask or directly produce enhanced speech. This design reduces unnecessary computation along the time dimension while retaining temporal information relevant to distinguishing speech from interference. The architecture is applicable not only to speech denoising but also to related enhancement problems such as echo suppression, dereverberation, bandwidth expansion, wind-noise removal, and recovery of missing speech. Rather than reporting a specific benchmark result, the patent focuses on the architectural principle and its main benefit: enabling effective neural speech enhancement with substantially lower computational requirements, making such processing more practical for mobile, wearable, and other resource-constrained devices.
