Acoustic Probing for Estimating the Storage Time and Firmness of Tomatoes and Mandarin Oranges

This paper introduces an acoustic probing technique to estimate the storage time and firmness of fruits; we emit an acoustic signal to fruit from a small speaker and capture the reflected signal with a tiny microphone. We collect reflected signals for fruits with various storage times and firmness conditions, using them to train regressors for estimation. To evaluate the feasibility of our acoustic probing, we performed experiments; we prepared 162 tomatoes and 153 mandarin oranges, collected their reflected signals using our developed device and measured their firmness with a fruit firmness tester, for a period of 35 days for tomatoes and 60 days for mandarin oranges. We performed cross validation by using this data set. The average estimation errors of storage time and firmness for tomatoes were 0.89 days and 9.47 g/mm2. Those for mandarin oranges were 1.67 days and 15.67 g/mm2. The estimation of storage time was sufficiently accurate for casual users to select fruits in their favorite condition at home. In the experiments, we tested four different acoustic probes and found that sweep signals provide highly accurate estimation results.

💡 Research Summary

The paper presents a low‑cost, non‑invasive acoustic probing method for estimating both the storage time and firmness of fruits, targeting everyday consumers rather than producers or distributors. The authors built a compact device (65 mm × 65 mm × 45 mm) that houses a small ear‑phone‑type speaker (SONY MDR‑EX150) and a silicon microphone (Knowles SPU0414HR5H‑SB) with an amplifier. The device emits acoustic probes and records the reflected signal at a 48 kHz sampling rate, with the microphone positioned roughly 5 mm from the fruit surface.

Four probe signals were evaluated: a 5 kHz single‑tone, a multi‑tone consisting of 3.5 kHz, 5 kHz and 6.5 kHz components, a linear sweep from 100 Hz to 10 kHz, and an exponential sweep covering the same range. Each probe lasts one second, and the four probes are concatenated into a single audio file to streamline data acquisition.

During data collection, 162 tomatoes and 153 mandarin oranges were harvested under controlled greenhouse conditions and stored at 8–9 °C in darkness. Tomatoes were measured every two days for 35 days, mandarin oranges every three days for 60 days. For each fruit, four spatial points were probed, yielding 648 sound recordings and firmness measurements for tomatoes and 612 for mandarin oranges. Firmness was measured with a digital fruit firmness tester (LUTORON FR‑5120) using a 3 mm tip; tomato skins were peeled to eliminate skin resistance.

Signal processing begins with a spectrogram of the raw recording; template matching identifies the start of each probe, after which a one‑second segment containing the probe is extracted. Two feature representations are derived from each segment: (1) a magnitude spectrum obtained via FFT with a Hamming window, and (2) 20‑dimensional Mel‑Frequency Cepstral Coefficients (MFCCs) computed with a 1024‑sample window. These features serve as inputs to regression models.

Two regression algorithms were tested: Support Vector Regression (SVR) and Gradient Boosting Regression (GBR). A three‑fold cross‑validation (k = 3) was performed, rotating the training and test splits. The results show that spectrum‑based features combined with sweep probes (both linear and exponential) achieve the highest accuracy. MFCCs and single‑tone probes performed noticeably worse, indicating that the broadband frequency response captured by sweeps better reflects internal fruit properties.

Quantitatively, the average absolute error in estimated storage time was 0.89 days for tomatoes and 1.67 days for mandarin oranges. Firmness estimation errors were 9.47 g/mm² for tomatoes and 15.67 g/mm² for mandarin oranges. The storage‑time errors are well within a one‑day margin, sufficient for consumers to decide whether a fruit meets their freshness preference. Firmness errors, while larger, still provide a useful coarse indicator of ripeness.

The study’s contributions are threefold: (1) introduction of an inexpensive acoustic probing system that requires only a speaker and microphone, (2) systematic evaluation of four probe types, demonstrating the superiority of sweep signals, and (3) validation on a relatively large dataset of two fruit types with cross‑validation, showing practical feasibility for consumer‑level freshness assessment.

Limitations include the focus on only two fruit varieties, the controlled temperature/humidity environment that may not reflect real‑world kitchen conditions, and potential sensitivity of the reflected signal to surface contaminants (soil, moisture) or fruit shape variations. Future work should expand the fruit repertoire, assess robustness across varying ambient conditions, and integrate the method into portable platforms such as smartphones or smart kitchen appliances. Additionally, developing calibration or compensation techniques for surface effects would enhance reliability for everyday use.