SONNET : Enhancing Time Delay Estimation by Leveraging Simulated Audio
Tegler, Erik LU ; Oskarsson, Magnus LU
and Åström, Kalle
LU
(2025)
27th International Conference on Pattern Recognition, ICPR 2024
In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
15320 LNCS.
p.289-303
- Abstract
Time delay estimation or Time-Difference-Of-Arrival estimates is a critical component for multiple localization applications such as multilateration, direction of arrival, and self-calibration. The task is to estimate the time difference between a signal arriving at two different sensors. For the audio sensor modality, most current systems are based on classical methods such as the Generalized Cross-Correlation Phase Transform (GCC-PHAT) method. In this paper we demonstrate that learning based methods can— even based on synthetic data—significantly outperform GCC-PHAT on novel real world data. To overcome the lack of data with ground truth for the task, we train our model on a simulated dataset which is sufficiently large and varied,... (More)
Time delay estimation or Time-Difference-Of-Arrival estimates is a critical component for multiple localization applications such as multilateration, direction of arrival, and self-calibration. The task is to estimate the time difference between a signal arriving at two different sensors. For the audio sensor modality, most current systems are based on classical methods such as the Generalized Cross-Correlation Phase Transform (GCC-PHAT) method. In this paper we demonstrate that learning based methods can— even based on synthetic data—significantly outperform GCC-PHAT on novel real world data. To overcome the lack of data with ground truth for the task, we train our model on a simulated dataset which is sufficiently large and varied, and that captures the relevant characteristics of the real world problem. We provide our trained model, SONNET (Simulation Optimized Neural Network Estimator of Timeshifts), which is runnable in real-time and works on novel data out of the box for many real data applications, i.e. without re-training. We further demonstrate greatly improved performance on the downstream task of self-calibration when using our model compared to classical methods.
(Less)
- author
- Tegler, Erik
LU
; Oskarsson, Magnus
LU
and Åström, Kalle
LU
- organization
- publishing date
- 2025
- type
- Chapter in Book/Report/Conference proceeding
- publication status
- published
- subject
- keywords
- Audio, Data Simulation, Generalized Cross-Correlation, Time Delay Estimation, Time-Difference-of-Arrival
- host publication
- Pattern Recognition - 27th International Conference, ICPR 2024, Proceedings
- series title
- Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
- editor
- Antonacopoulos, Apostolos ; Chaudhuri, Subhasis ; Chellappa, Rama ; Liu, Cheng-Lin ; Bhattacharya, Saumik and Pal, Umapada
- volume
- 15320 LNCS
- pages
- 15 pages
- publisher
- Springer
- conference name
- 27th International Conference on Pattern Recognition, ICPR 2024
- conference location
- Kolkata, India
- conference dates
- 2024-12-01 - 2024-12-05
- external identifiers
-
- scopus:85212247795
- ISSN
- 1611-3349
- 0302-9743
- ISBN
- 9783031784972
- DOI
- 10.1007/978-3-031-78498-9_20
- language
- English
- LU publication?
- yes
- additional info
- Publisher Copyright: © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.
- id
- aa3af396-6cc5-40bf-8e28-18bce4ce6e6e
- date added to LUP
- 2025-01-22 11:36:01
- date last changed
- 2025-07-10 01:17:58
@inproceedings{aa3af396-6cc5-40bf-8e28-18bce4ce6e6e,
abstract = {{<p>Time delay estimation or Time-Difference-Of-Arrival estimates is a critical component for multiple localization applications such as multilateration, direction of arrival, and self-calibration. The task is to estimate the time difference between a signal arriving at two different sensors. For the audio sensor modality, most current systems are based on classical methods such as the Generalized Cross-Correlation Phase Transform (GCC-PHAT) method. In this paper we demonstrate that learning based methods can— even based on synthetic data—significantly outperform GCC-PHAT on novel real world data. To overcome the lack of data with ground truth for the task, we train our model on a simulated dataset which is sufficiently large and varied, and that captures the relevant characteristics of the real world problem. We provide our trained model, SONNET (Simulation Optimized Neural Network Estimator of Timeshifts), which is runnable in real-time and works on novel data out of the box for many real data applications, i.e. without re-training. We further demonstrate greatly improved performance on the downstream task of self-calibration when using our model compared to classical methods.</p>}},
author = {{Tegler, Erik and Oskarsson, Magnus and Åström, Kalle}},
booktitle = {{Pattern Recognition - 27th International Conference, ICPR 2024, Proceedings}},
editor = {{Antonacopoulos, Apostolos and Chaudhuri, Subhasis and Chellappa, Rama and Liu, Cheng-Lin and Bhattacharya, Saumik and Pal, Umapada}},
isbn = {{9783031784972}},
issn = {{1611-3349}},
keywords = {{Audio; Data Simulation; Generalized Cross-Correlation; Time Delay Estimation; Time-Difference-of-Arrival}},
language = {{eng}},
pages = {{289--303}},
publisher = {{Springer}},
series = {{Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)}},
title = {{SONNET : Enhancing Time Delay Estimation by Leveraging Simulated Audio}},
url = {{http://dx.doi.org/10.1007/978-3-031-78498-9_20}},
doi = {{10.1007/978-3-031-78498-9_20}},
volume = {{15320 LNCS}},
year = {{2025}},
}