Tomasz Grzywalski

Mr. Tomasz Grzywalski

StethoMe, Poznań, Wielkopolskie, Poland
orcid

Research Focus & Metrics

Academic Career Length

8Years (2019-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
4 March 2025 in Poultry Science
Verified
Journal Article
Open access
PSPatricia de Carvalho Soster
Patricia de Carvalho Soster
Tomasz GrzywalskiTomasz Grzywalski
Tomasz Grzywalski
YUANBO HOUYUANBO HOU
YUANBO HOU
PTPieter Thomas
Pieter Thomas
ADAnnelike Dedeurwaerder
Annelike Dedeurwaerder
GAGunther Antonissen
Gunther Antonissen
The poultry industry relies on highly efficient production systems. For sustainable food production, where maintaining broiler welfare is crucial, it is essential to have robust data collection systems and automated methods for assessing broiler health and welfare. This paper presents the development and implementation of an acoustic system designed to detect and differentiate between four distinct vocalizations of broiler chickens-pleasure notes, distress calls, short peeps, and warbles-while filtering out background noise and other vocalizations. The vocalization detector is designed as a convolutional neural network with 11 two-dimensional convolutional layers and one one-dimensional convolutional layer. For training, a manually labeled vocalization library was built (>2k samples, with a total duration of 190 minutes), based on a large set of continuous audio recordings of ten male Ross 308 broiler chicks aged from 1 to 36 days. An extension with a subset of the AudioSet dataset was made to include background sounds. With this library, an overall balanced accuracy of 91.1 % was achieved by the neural network-based recognizer.

14 May 2025 in Poultry Science
Verified
Journal Article
Open access
PSPatricia Soster Soster
Patricia Soster Soster
Tomasz GrzywalskiTomasz Grzywalski
Tomasz Grzywalski
KBKobe Buyse
Kobe Buyse
PTPieter Thomas
Pieter Thomas
Camila Lopes CarvalhoCamila Lopes Carvalho
Camila Lopes Carvalho
GAGunther Antonissen
Gunther Antonissen
Sound analysis of animal vocalizations may provide valuable insights into their emotional state and behavioral responses to environmental stimuli. In broiler chickens, four primary vocalization types-distress calls (DC), short peeps (SP), warbles (W), and pleasure notes (PN)-are well-characterized through sound analysis. The objective of this study was to identify the normal vocalization patterns of broiler chickens throughout the on-farm stage, considering different times of the day (morning, midday, afternoon, and night), and to assess the effects of heat stress (HS) and multifunctional elevated platforms (MP) enrichment on these patterns. The trial included 1,680 male Ross 308 chickens, housed under either thermoneutral or HS conditions during the final two weeks, with half the pens enriched with MP. DC, SP, PN, and W were automatically monitored using custom-build broiler vocalization recognizer. When a vocalization did not fit one of the four categories, it was classified as another type of vocalization (OV). Time of day significantly influenced SP, DC, W, and PN. Throughout the starter, grower, and finisher phases, all treatments exhibited similar vocalization patterns across age and time of the day, highlighting the strong influence of daily periods on vocalizations. SP had the highest frequency in the evening and the lowest at night. DC peaked at midday, while W reached its highest frequency at night and was least frequent at midday. PN were most common at night but significantly lower during midday and evening. SP and DC were the most prevalent vocalizations, while PN occurred less frequently and were most prominent in broilers of 1 week old. W remained consistently low throughout the broilers' lifespan. DC, SP, and OV increased with age, whereas PN declined, indicating a shift and more diverse vocal communication patterns as broilers mature. The presence of MP and HS did not influence vocalization patterns. These results demonstrate significant changes in vocalization patterns with age and time of the day throughout the on-farm cycle, while remaining relatively unaffected by the presence of MP or HS.

18 February 2025 in Journal of Cystic Fibrosis
Verified
Journal Article
Closed access
Henryk MazurekHenryk Mazurek
Henryk Mazurek
AEAndrzej Emeryk
Andrzej Emeryk
KJKamil Janeczek
Kamil Janeczek
EDEric Derom
Eric Derom
BKBarbara Kamińska
Barbara Kamińska
JKJęDrzej Kociński
JęDrzej Kociński

22 March 2022 in Sensors
Verified
Journal Article
Open access
PDF file
Tomasz GrzywalskiTomasz Grzywalski
Tomasz Grzywalski
SDSzymon Drgas
Szymon Drgas
Monaural speech enhancement aims to remove background noise from an audio recording containing speech in order to improve its clarity and intelligibility. Currently, the most successful solutions for speech enhancement use deep neural networks. In a typical setting, such neural networks process the noisy input signal once and produces a single enhanced signal. However, it was recently shown that a U-Net-based network can be trained in such a way that allows it to process the same input signal multiple times in order to enhance the speech even further. Unfortunately, this was tested only for two-iteration enhancement. In the current research, we extend previous efforts and demonstrate how the multi-forward-pass speech enhancement can be successfully applied to other architectures, namely the ResBLSTM and Transformer-Net. Moreover, we test the three architectures with up to five iterations, thus identifying the method’s limit in terms of performance gain. In our experiments, we used the audio samples from the WSJ0, Noisex-92, and DCASE datasets and measured speech enhancement quality using SI-SDR, STOI, and PESQ. The results show that performing speech enhancement up to five times still brings improvements to speech intelligibility, but the gain becomes smaller with each iteration. Nevertheless, performing five iterations instead of two gives additional a 0.6 dB SI-SDR and four-percentage-point STOI gain. However, these increments are not equal between different architectures, and the U-Net and Transformer-Net benefit more from multi-forward pass compared to ResBLSTM.

21 March 2026 in Poultry Science
Verified
Journal Article
Open access
PSPatricia Soster Soster
Patricia Soster Soster
Tomasz GrzywalskiTomasz Grzywalski
Tomasz Grzywalski
Camila Lopes CarvalhoCamila Lopes Carvalho
Camila Lopes Carvalho
IKImad Khan
Imad Khan
PTPieter Thomas
Pieter Thomas
GAGunther Antonissen
Gunther Antonissen
Broiler vocalizations may contain relevant information about the welfare and health status; however, there is limited information on how disease affects vocal patterns. This study aimed 1) to investigate whether specific vocalization changes emerge in response to two important poultry diseases: intestinal (Eimeria spp.) and respiratory (infectious bronchitis virus, IBV; combined with avian pathogenic Escherichia coli, APEC), and 2) to characterize changes in broiler chicken vocalizations over age (14-41 days of age), and across time of the day (morning: 6:00 AM-12:00 PM; afternoon: 12:00 PM-6:00 PM; evening: 6:00 PM-12:00 AM; night: 12:00 AM-6:00 AM). Vocalizations were monitored using an automated AI-based broiler vocalization detector previously developed by our group. A total of 176 Ross 308 broilers were included across four experimental rounds, each consisting of three treatments (control and the two disease challenges) with four replicates each of 11 birds per group. Groups were housed in 3 m² pens and continuously recorded from day 14 to day 41 using centrally mounted microphones. Disease challenges were administered on day 14 (Eimeria spp.) and day 21 (IBV followed by APEC on day 24). Audio recordings were analyzed using a convolutional neural network trained to detect five vocalization categories: distress calls (DC), short peeps (SP), pleasure notes (PN), warbles (W), and other vocalizations (OV). Vocalizations were aggregated hourly to be evaluated across treatments, age and time of the day. Statistical analyses were performed using mixed-effects models in SAS (PROC MIXED). No significant effects of treatment were detected for all vocalization types (p > 0.05). However, significant age-related changes in vocal expression were observed (p < 0.05), with DC, SP, and PN decreasing progressively over age, while OV increased. Also, DC, SP, and OV were less frequent during the night from 14 to 28 days of age when compared to the other times of the day (p < 0.05), indicating that vocal expression was primarily driven by age- and diurnal period-related changes rather than experimental disease challenge. These results highlight both the potential of vocalization patterns baselines to better understand broilers vocal behavior and current limitations of acoustic monitoring for detecting health disturbances in broiler chickens.

4 September 2025 in Poultry Science
Verified
Journal Article
Open access
Tomasz GrzywalskiTomasz Grzywalski
Tomasz Grzywalski
PSPatricia Soster Soster
Patricia Soster Soster
YUANBO HOUYUANBO HOU
YUANBO HOU
PTPieter Thomas
Pieter Thomas
Frank TuyttensFrank Tuyttens
Frank Tuyttens
DBDick Botteldooren
Dick Botteldooren
This study investigates and characterizes vocalizations in broiler chickens from 1 to 35 days of age using advanced acoustic analysis and machine learning techniques. Understanding broiler behavior, particularly vocalizations, is crucial for improving animal welfare in both on-farm and laboratory conditions. While four known vocalizations (distress calls, short peeps, warbles, and pleasure notes) are well documented using sound analysis, there remains a gap in understanding the full vocal repertoire of broiler chickens, which may hold key insights into their emotional and physiological states. Using a deep learning-based vocalization recognizer and recursive clustering algorithms, we identified 42 distinct sound clusters - in addition to the 4 known vocalizations - from recordings of healthy broiler chickens, eventually narrowing them down to 10 key clusters that potentially represent novel vocalizations. These vocalizations were analyzed for their frequency, duration, and acoustical power, and their temporal distribution was examined. The findings suggest that broilers expand their vocal repertoire as they age, presenting a more diverse repertoire in the later stages of life. Despite the limited sample size and absence of statistical replicates, this study offers valuable insights into the complexity of broiler vocalizations. This research contributes to the growing body of knowledge on broiler auditory communication and opens new possibilities for automated vocalization monitoring in chicken farming.