This article discusses the innovative application of a Cartesian robot manipulator with acoustic feedback for calibration and precise positioning of a string-excitation element in investigating stringed instruments. It describes an experiment in which an acoustic guitar string is automatically excited with different guitar picks. The robot’s end effector positioning system utilizes limit switches, acting as a mechanical sensor, which provides feedback to the linear actuators that are equipped with stepper motors. The authors detail the research challenges they faced and propose a positioning algorithm that makes use of a microphone as an acoustic sensor, improving the calibration of the end effector’s position.

Celem pracy jest zbadanie zasobu leksykalnego używanego do opisu barwy dźwięku w języku polskim. W przeciwieństwie do wysokości, głośności czy długości trwania dźwięku barwa nie posiada oczywistych korelatów fizycznych, a co za tym idzie – precyzyjnych określeń językowych; do jej opisu są stosowane głównie metafory. W ramach pracy przeprowadzono eksperyment eksploracyjny, w którym badani podawali jak największą liczbę jednostek opisujących dźwięki instrumentów muzycznych. Na tej podstawie wybrano określenia dotyczące barwy dźwięku, zaproponowano podział uzyskanych słów na kategorie oraz porównano uzyskany zbiór z zasobem określeń przywoływanych w literaturze. Uzyskane odpowiedzi w dużym stopniu pokrywają się z wynikami wcześniejszych analiz, niemniej jednak niektóre popularne specjalistyczne terminy nie pojawiły się w odpowiedziach uczestników przeprowadzonego badania.

Hallucinations of deep neural models are amongst key challenges in automatic speech recognition (ASR). In this paper, we investigate hallucinations of the Whisper ASR model induced by non-speech audio segments present during inference. By inducting hallucinations with various types of sounds, we show that there exists a set of hallucinations that appear frequently. We then study hallucinations caused by the augmentation of speech with such sounds. Finally, we describe the creation of a bag of hallucinations (BoH) that allows to remove the effect of hallucinations through the post-processing of text transcriptions. The results of our experiments show that such post-processing is capable of reducing word error rate (WER) and acts as a good safeguard against problematic hallucinations.

This study presents a comparative performance analysis of three sensor technologies—microphone, magnetic pickup, and laser Doppler vibrometer—for capturing string vibration under varied excitation conditions: striking, plectrum plucking, and wire plucking. Two different magnetic pickups are included in the comparison. Measurements were taken at multiple excitation levels on a simplified electric guitar mounted on a stable platform with repeatable excitation mechanisms. The analysis focuses on each sensor’s capacity to resolve fine-scale waveform features during the initial attack while also taking into account its capability to measure general changes in instrument dynamics and timbre. We evaluate their ability to distinguish vibro-acoustic phenomena resulting from changes in excitation method and strength as well as measurement location. Our findings highlight the significant influence of sensor choice on observable string vibration. While the microphone captures the overall radiated sound, it lacks the required spatial selectivity and offers poor SNR performance 34 dB lower then other methods. Magnetic pickups enable precise string-specific measurements, offering a compelling balance of accuracy and cost-effectiveness. Results show that their low-pass frequency characteristic limits temporal fidelity and must be accounted for when analysing general sound timbre. Laser Doppler vibrometers provide superior micro-temporal fidelity, which can have critical implications for physical modeling, instrument design, and advanced audio signal processing, but have severe practical limitations. Critically, we demonstrate that the required optical target, even when weighing as little as 0.1% of the string’s mass, alters the string’s vibratory characteristics by influencing RMS energy and spectral content.