Unveiling the Aromas and Sensory Evaluation of Hakko Sobacha: A New Functional Non-Dairy Probiotic Fermented Drink

At the dawn of a food transition encouraging the consumption of healthy and sustainable non-dairy probiotic products, the development of a fermented functional drink based on Sobacha is considered. Sobacha is an infusion of roasted buckwheat seeds widely consumed in Asian countries for its health benefits. As fermentation improves the nutritional and organoleptic status of grains, the mixed fermentation process involved in the development of kombucha beverages (fermented sweet tea) is conducted by inoculating a symbiotic culture of bacteria and yeasts into the transposable matrix (Sobacha instead of tea). Sobacha, a healthy pseudo-cereal matrix with promising aromas, could be fermented to potentially develop an innovative drink, named “Hakko Sobacha”. This neologism would reveal the fermented character of the infusion, Hakko meaning fermented in Japanese. Considering the beverage characterization, the kinetics of the volatile organic compound syntheses were determined using stir-bar sorptive extraction followed by gas chromatography coupled to mass spectrometry analysis. Odor-active compounds were theoretically calculated to estimate the flavor composition. Finally, sensory analyses highlighted the appreciation and preferences of the consumer towards the beverages. The fermentative yield differences observed between the two buckwheat concentration modalities tested seemed to be correlated with the sugar and nutrient levels available from the starch (buckwheat) matrix. Having characterized Hakko Sobacha, this study proposed the possibility of developing new beverages by monitoring the fermentative process. This should enable improved control and enhancement of their sensorial properties, which could in turn lead to greater customer acceptability.


Introduction
There is a growing demand for the development of functional foods in the market, i.e., satisfying hunger and nutrient intake, but also preventing chronic diseases and improving physical and mental health. An example of a widely consumed food product is probiotics (provision of beneficial microorganisms (MOs) in the gut microbiota) [1,2]. Thus, in addition to stimulating the growth of beneficial microflora, they have direct effects on both satiety and the intestinal barrier [3]. Moreover, the probiotic market is growing rapidly, opening new opportunities for non-dairy substrates. Thus, pseudo-cereal substrates are promising candidates for the growth of lactobacilli and bifidobacterial strains [4]. Kombucha (from the Japanese "kombu", algae and "cha", tea) has gained immense popularity in recent years due to its many health benefits [5]. This fermented drink presents the characteristics of a functional food [6]. Originating in China, Korea, and Japan (220 BC), it is popular • Does Sobacha fermentation induce the acidification of the medium towards a pH range allowing the inhibition of pathogenic microorganisms? • What are the production kinetics of the molecules of interest in the degradation of fermentable sugars? Are the levels of ethanol produced at the end of fermentation in line with the standards for designating non-alcoholic beverages? • What are the major volatile organic compounds developed during Sobacha fermentation? Among these, which aromas contribute significantly to the product's final aromatic profile? • From a sensory point of view, what are the consumer's opinions and preferences regarding the beverages formulated?
To address these issues, after analyzing the physico-chemical parameters of the formulated beverage as a function of the fermentation key parameters (such as the fermentation time), the kinetics of the volatile organic compound syntheses were determined using stir-bar sorptive extraction followed by analysis with gas chromatography coupled to mass spectrometry (SBSE-GC-MS); SBSE allows for then optimal capture of volatile and semi-volatile compounds in aqueous matrices. The correlation with microbial metabolic pathways was then established, ensuring a constant quality and stability of the produced metabolites regarding the Sobacha substrate.
Although kombucha presents a complex mixture of VOCs, only a portion of them contribute to the overall aroma and influence consumer perception. Ortho-nasal detection odor thresholds were then considered to determine odor-active compounds and evaluate theoretical sensory profiles of Hakko Sobacha. Finally, the aromatic compound kinetics during the fermentation process were coupled with sensory analysis. This methodology will in time allow for the design of new drinks by controlling the fermentation process, as well as meet consumer preferences.

Results and Discussion
2.1. Physico-Chemical Characterization of Hakko Sobacha during Fermentation 2.1.1. Evolution of pH during Fermentation As illustrated in Figure 1, the pH values of Hakko Sobacha exponentially decreased during the first aerobic fermentation. The pH was revealed to be non-significantly different between the first and second fermentation days for both kasha concentrations considered (means of 5.76 ± 0.05 and 5.71 ± 0.05 g/L at D0 to 4.55 ± 0.07 to 4.68 ± 0.05 g/L at D2, for 10 and 50 g/L kasha concentration modalities, respectively). From the third day (D3) until the end of F1, the pH decrease started to be significantly different between the two kasha concentration modalities and fermentation times, with higher pH values for the fermentation medium with the lower kasha concentration (means of 4.29 ± 0.06 and 3.82 ± 0.09 g/L (p < 0.01) at D3 to 3.30 ± 0.03 and 3.18 ± 0.01 g/L (p < 0.01) at D10, for 10 and 50 g/L kasha concentration modalities, respectively). A significant increase in pH occurred during the second anaerobic fermentation, reaching mean values of 3.35 ± 0.02 and 3.26 ± 0.01 g/L (p < 0.01) at D25 (for 10 and 50 g/L kasha concentration modalities, respectively).  When comparing Hakko Sobacha and kombucha pH evolution trends, similar behaviors should be observed, namely an exponential decrease followed by a slower decrease, as often mentioned in the literature for the kombucha fermentation process. Moreover, studies focusing on the anaerobic stage of the kombucha fermentation process also observed a slight pH increase over time. However, given the lower polyphenol content in the buckwheat infusion compared to the tea matrix, the significant difference in pH observed at the initial time could be linked to the lower acidity of the infusions (initial pH of 4.05 for a black/green tea kombucha).
Therefore, the pH decrease may reflect the acidification of both fermentation media and thus the microorganism's ability to metabolize diverse products in terms of Sobacha composition. Since this acidification is accentuated when the kasha concentration is higher, the hypothesis that the presence of an environment richer in carbon sources and nutrients available for metabolic activity, such as in buckwheat, could be advanced.

Evolution of SCOBY Production during Fermentation
Microbial cellulose growth is a direct parameter for assessing the microorganism's evolution in the fermentation medium. More precisely, it is directly linked to the ability of most Acetobacter and Gluconobacter genera to synthetize cellulose from available carbon sources. Subsequently, the evolution of SCOBY weights as a function of kasha concentration is presented in Figure 2.  From an initial SCOBY inoculum of ±5 g for both concentration modalities, the subsequent nascent SCOBY pellicle weights increased significantly during the aerobic fermentation process. Indeed, significant differences were observed for the resulting cellulose growth as a function of the buckwheat concentration (wet weight means of 1.05 ± 0.26 and 12.47 ± 2.80 g (p < 0.01) for 10 and 50 g/L kasha concentration modalities, respectively).
Concerning dry matter, very high significant differences were reported (dry weight means of 0.23 ± 0.13 and 7.49 ± 1.37 g (p < 0.001) for 10 and 50 g/L kasha concentration modalities, respectively). This could be related to the absorption capacity of the cellulose films generated (closely linked to the thickness of the biofilms), allowing liquid to be retained to a greater or lesser extent.
These observations are in accordance with the different acidification profiles obtained from kasha concentration modalities (Section 2.1.1.) and the carbon sources available for cellulose production during the aerobic fermentation step. Indeed, the presence of a higher simple sugar content when the buckwheat concentration is increased allows for both the microorganism's metabolic activity, which is responsible for cellulose synthesis, and acidification to be improved.

Carbohydrate, Alcohol, and Organic Acid Concentration Kinetics during Hakko Sobacha Fermentation
The evolution of carbohydrates, organic acids, and ethanol contents as a function of buckwheat concentration and fermentation time is illustrated in Figure 3.
Considering the carbohydrate analysis, the initial sucrose contents decreased over time. From the initial stage to D3, no significant difference was observed regarding the kasha concentration (means of 62.36 ± 0.38 and 61.50 ± 0.50 g/L at D0, 55.19 ± 0.15 and 50.08 ± 0.63 g/L at D3, for 10 and 50 g/L kasha concentration modalities, respectively). From D6, the sucrose contents showed significant differences from the kasha concentrations (means of 53.69 ± 0.15 and 41.87 ± 0.32 g/L (p < 0.01) at D6, to end with 26.37 ± 2.65 and 3.02 ± 1.62 g/L (p < 0.001) at D25, for 10 and 50 g/L kasha concentration modalities, respectively). On the other hand, the amounts of residual sucrose reported in the literature for kombucha vary according to the initial sucrose input. The evolution of carbohydrates, organic acids, and ethanol contents as a function of buckwheat concentration and fermentation time is illustrated in Figure 3. Considering the carbohydrate analysis, the initial sucrose contents decreased over time. From the initial stage to D3, no significant difference was observed regarding the kasha concentration (means of 62.36 ± 0.38 and 61.50 ± 0.50 g/L at D0, 55.19 ± 0.15 and 50.08 ± 0.63 g/L at D3, for 10 and 50 g/L kasha concentration modalities, respectively). From D6, the sucrose contents showed significant differences from the kasha concentrations (means of 53.69 ± 0.15 and 41.87 ± 0.32 g/L (p < 0.01) at D6, to end with 26.37 ± 2.65 and 3.02 ± 1.62 g/L (p < 0.001) at D25, for 10 and 50 g/L kasha concentration modalities, The buckwheat concentration therefore seemed to have a great impact on the metabolic activity of sucrose conversion by the microorganisms during the first fermentation. In addition, the significant difference in sugar levels observed at the end of F2 as a function of buckwheat concentration could be explained by the modification of the fermentation medium. Indeed, the microbial consortium could benefit from a larger amount of nutrients present in the infusion and subsequently released during fermentation. In other words, the starch and complex buckwheat sugars released in the infusion and partially hydrolyzed by the microbiome represent another source of carbohydrates, enabling them to further acidify the medium.
The degradation kinetics of sucrose into glucose and fructose units by yeast taxa via the glycolysis pathway ensures that simple carbohydrates are assimilated by the latter. Consequently, as the simple sugar level is inversely proportional to that of sucrose, it increases during fermentation [2]. Indeed, the glucose and fructose contents increased significantly over time, with a higher concentration of fructose at all points (except for the second fermentation). Moreover, the medium exhibited higher levels of simple sugars when the buckwheat concentration in the fermentation medium was higher. The initial glucose levels were 0.19 ± 0.02 and 0.26 ± 0.03 g/L (p < 0.05), and the final levels reached 10.22 ± 0.73 and 13.00 ± 0.44 g/L (p < 0.01) for the 10 and 50 g/L kasha concentration modalities respectively.
During the hydrolysis of sucrose, yeast concurrently converts both fructose and glucose into ethanol and CO 2 on one side, with a more rapid utilization of glucose compared to fructose. Additionally, the preferential carbon source for AAB is glucose, leading to the synthesis of microbial cellulose. Since Sobacha is a food matrix that is rich in starch and made up of glucose units, the main source of fructose in the fermentation medium is sucrose.
The ethanol levels significantly increased as a function of kasha concentration during the fermentation process. Starting from 0.04 ± 0.01% and 0.10 ± 0.01% (p < 0.01) at D0 to 0.05 ± 0.02% and 0.19 ± 0.02% (p < 0.01) at D10 during the aerobic step, ending with ethanol levels of 0.32 ± 0.03% and 0.89 ± 0.01% (p < 0,0001) at D25 for the anaerobic process (for 10 and 50 g/L kasha concentration modalities, respectively). Ethanol synthesis at the beginning of fermentation is indeed reported in the literature and correlated with an exponential increase in the yeast population. Moreover, studies reported a kombucha final ethanol content of approximately 0.5% and 0.6% after 12 and 20 days of fermentation, respectively (for initial sucrose contents of 50 and 100 g/L, respectively), which is in between the results for Sobacha.
Yeasts drive ethanol production thanks to the availability of carbonaceous and nitrogenous substrates as well as oxygen in the medium. A further increase in ethanol occurs during the anaerobic fermentation step, as yeasts become the dominant microorganism in the medium, to the detriment of bacteria. The availability of additional glucose from starch hydrolysis during the anaerobic step allows for obtaining a higher final ethanol content.
At the end of the anaerobic fermentation, both beverage's ethanol levels were below the 1.2% (v/v) considered by the European regulation (1169/2011) for characterizing nonalcoholic drinks.
Furthermore, due to the presence of a symbiotic ecosystem in Hakko Sobacha, the ethanol produced by the yeasts was then converted via multiple metabolic pathways into organic acids by different bacteria genera. In this way, the acetic acid contents increased progressively over time, characterized as significantly different from D3 as a function of kasha concentration. The acetic acid levels started from 0.06 ± 0.01 and 0.07 ± 0.01 g/L (p > 0.05) at D0 until 4.49 ± 0.13 and 7.13 ± 0.12 g/L (p < 0.001) at D25 (for 10 and 50 g/L kasha concentration modalities, respectively).
Previous studies focused on traditional kombucha reported a significant initiation of acetic acid production at D3 (reaching approximatively 5 g/L for an initial sucrose level of 100 g/L) and an exponential increase in bacteria population growth from this stage.
Despite the low concentrations of lactic acid measured during the fermentation process, they increased significantly over time starting from D3, depending on the kasha concentration. The lactic acid production started at 0.02 ± 0.01 and 0.07 ± 0.01 g/L (p < 0.01) at D3 to reach final values of 0.15 ± 0.03 and 0.33 ± 0.03 g/L (p < 0.01) at D25 (for 10 and 50 g/L kasha concentration modalities, respectively). Indeed, LAB represent a minor part of the microbial consortium present in the kombucha SCOBY compared to AAB; lactic acid is then less produced.  Data are expressed as the mean ± SD; "nd": Not detected. RI: Retention indices on HP-5MS column were determined by n-alkanes. RI lit: Literature retention indices reported for non-polar capillary HP-5MS column.  Data are expressed as the mean ± SD; "nd": Not detected. RI: Retention indices on HP-5MS column were determined by n-alkanes. RI lit: Literature retention indices reported for non-polar capillary HP-5MS column.
The Sobacha matrix before fermentation-considering all kasha concentration modalitieswas mainly composed of carboxylic acids, terpenes, and pyrazines.
The buckwheat infusion presented high levels of palmitic (650.24 and 304.18 ppm) and palmitoleic acids (180.24 and 144.70 ppm, for 10 and 50 g/L kasha concentration modalities, respectively). In fact, fatty acids are a specific type of carboxylic acids in tea and infusion matrices. Another major class from plant-based infusions are terpenes, presenting properties involved in the plant chemical defense. The major Sobacha terpenes were p-cymene (676.32 and 510.21 ppm) and m-myrcene (510.21 and 439.92 ppm, for 10 and 50 g/L kasha concentration modalities, respectively). Lastly, pyrazines present in the infusion are for the most part related to the buckwheat seed roasting process. Thus, pyrazine levels are positively correlated with kasha concentration. The major pyrazine representatives present in the Sobacha were 2-methoxy-5-methylpyrazine (342.21 ppm) and 2-ethylpyrazine (129.11 ppm) for the 50 g/L kasha concentration modality. Figure 4 graphically illustrates the average content changes of all compound classes as a function of buckwheat concentration and fermentation time, and a link between the VOC diversity profile and the total content of released compounds could be established during the fermentation process. In fact, fermentation dynamically changes the VOC profile of the starting infusion into a different mixture of compounds.
The classes present in the initial infusion tended to decrease during fermentation. As fermentation started, the microbial consortium activity assimilated the organic substrates present in the infusion media to synthetize metabolites, consequently leading to new VOC diversity and proportions.
As described by Suffys et al. (2023) [23], classes such as alcohols, ketones, aldehydes, and esters are the main metabolites synthesized by the microbiome in kombucha-type fermentation. Indeed, the related metabolic pathways could be similar and adapted to the nutrients present in the infusion.
Carboxylic acids are also major components produced during the fermentation process, followed by esters. The final average contents reported for carboxylic acids were 587.30 and 322.45 ppm, and for alcohols 30.91 and 146.94 ppm (for 10 and 50 g/L kasha concentration modalities, respectively). Indeed, it has been reported in previous studies on kombucha that these two classes of molecules constitute substrates for the synthesis of other types of compounds such as esters, phenols, and terpenes mainly (molecules formed using aldehydes and ketones).
Moreover, mixed fermentation allows to produce enriched VOC profiles by non-Saccharomyces yeast strains such as Zygosaccharomyces sp., Dekkera sp., Candida sp., and Pichia sp. They contribute to the final product aroma but also to the formation of organic acids by their oxidation. Principally, together with aldehydes, alcohols are the main substrates used by the MO present in the SCOBY to synthesize aromatic volatile acids that acidify the medium and esters.
Intermediary classes such as aldehydes and ketones are converted during fermentation into secondary metabolites and aromatic molecules. Indeed, their contents at the end of the fermentation process were relatively low (86.97 and 119.73 ppm for aldehydes, 15.64 and 28.16 ppm for ketones, for 10 and 50 g/L kasha concentration modalities, respectively).
Finally, focusing again on the VOC classes mainly formed during the fermentation process, the total ester contents reached 143.70 and 335.34 ppm, and for phenol 24.38 and 128.56 ppm, for the 10 and 50 g/L kasha concentrations, respectively. The major ester synthetized was 2-phenylethyl acetate (final contents of 120.67 ± 8.41 and 57.55 ± 2.20 ppm, for 10 and 50 g/L kasha level modalities, respectively). Developed by the acetic acid condensation reaction, this ester participates in kombucha-type fermentation. In this way, ethyl caprylate (95.52 ± 1.29 ppm at D25, for 50 g/L of kasha) was the major ester metabolized from the ethanol substrate. At last, 4-ethylguaiacol was the major phenol found in Hakko Sobacha (116.18 ± 3.33 ppm at D25, for 50 g/L of kasha), an aromatic phenol produced by the yeast genus Brettanomyces.
Molecules 2023, 28, x FOR PEER REVIEW 14 of 32 Figure 4 graphically illustrates the average content changes of all compound classes as a function of buckwheat concentration and fermentation time, and a link between the VOC diversity profile and the total content of released compounds could be established during the fermentation process. In fact, fermentation dynamically changes the VOC profile of the starting infusion into a different mixture of compounds.
The classes present in the initial infusion tended to decrease during fermentation. As fermentation started, the microbial consortium activity assimilated the organic substrates present in the infusion media to synthetize metabolites, consequently leading to new VOC diversity and proportions.
As described by Suffys et al. (2023) [23], classes such as alcohols, ketones, aldehydes, and esters are the main metabolites synthesized by the microbiome in kombucha-type fermentation. Indeed, the related metabolic pathways could be similar and adapted to the nutrients present in the infusion.  A PCA was performed to analyze the variation among the different VOC profiles during the fermentation process ( Figure 5). The first dimension explained 45.9% of the total variance, represented on the horizontal axis. A variable correlation plot illustrates the possible group formations, depending on their position on the 1:2 plan ( Figure 5A). First, the infusions-considering all kasha concentrations-were positively correlated together, inducing similar VOC profiles. The overall fermentation process differed according to the kasha concentration in the infusion, as the variables were situated in fourth quadrant (QIV) and first quadrant (QI) (for 10 and 50 g/L kasha, respectively). Moreover, the contribution scale indicated that the data set variability was further explained by the Hakko Sobacha beverage based on the 10 g/L kasha concentration modality. Finally, considering all kasha concentrations, the fermentation stage groups traducing similar VOC profiles were the following: D1-D2, D3-D7 and D8-D25. Moreover, the trends provided by the second dimension revealed no additional information considering this dataset (explanation of 13.7% of the total variability). To resume, a major significant change in the VOC profiles between the beginning (D1-D2), the intermediate (D3-D7), and the end (D8-D25) of the fermentation stages was highlighted by the PCA. Furthermore, as illustrated by the individual representations ( Figure 5B), the classes that explained the major variability between kasha concentrations and fermentation times were carboxylic acids, terpenes, aldehydes, and phenols (most represented along Dim1).

Theoretical Aromatic Characterization of Hakko Sobacha and Estimation of Sensory Profiles
As the microbial consortium plays a crucial role in the development of flavor quality during fermentation, the formulated beverages will now be analyzed from this point of view.

Theoretical Aromatic Characterization of Hakko Sobacha and Estimation of Sensory Profiles
As the microbial consortium plays a crucial role in the development of flavor quality during fermentation, the formulated beverages will now be analyzed from this point of view.
The evolution of aromatic molecules is presented in Tables 3 and 4 (for 10 and 50 g/L kasha concentration modalities, respectively). Based on the VOC levels and perception thresholds in water reported in the literature, odor-activity values were determined during the fermentation process. The odor-activity value (OAV) explains the contribution of a VOC to a specific aroma. When the OAV is greater than 1, it indicates that the considered compound has a significant contribution to the overall aroma in the corresponding food system. Indeed, the greater the OAV is, the higher the contribution of the VOC to the aroma is.
During the fermentation process, the Sobacha sensory profile was modified through MO activity, resulting in the development of diverse aromatic molecules with different perception profiles. The synthesis of 2-phenylethyl acetate (final OAV of 16.09 and 7.67 for 10 and 50 g/L kasha concentration modalities, respectively) contributed a pineapple aroma to the final beverage's flavor. Ethyl caprylate was another ester metabolized during fermentation, bringing sweet-waxy-fruity-pineapple notes (final OAV of 12.74 for the kasha concentration modality of 50 g/L). Pelargonic acid, giving waxy-dairy-cheesy aromas, also contributed to the final Hakko Sobacha flavor (final OAV of 7.51 for the kasha concentration modality of 10 g/L). However, even if the initial infusion constituents globally decreased across the fermentation stages, they still contributed to the aroma system as their OAV remained greater than 1. Table 3. OAV evolution of most representative volatile organic compounds (% area > 5%) with Hakko Sobacha (kasha concentration 10 g/L) fermentation time (day) conducted at 25 • C (OAV = concentration/perception threshold).    The correlation between the changes in VOC profiles during the fermentation process and metabolic pathways is reported in the literature. Therefore, aroma-active compounds also fluctuate. Considering the metabolic pathways employed in kombucha-type fermentation, multiple aromatic compounds are referred to as microbial. However, although the beverage s final flavor maintains common compounds that contributes to the kombucha- The correlation between the changes in VOC profiles during the fermentation process and metabolic pathways is reported in the literature. Therefore, aroma-active compounds also fluctuate. Considering the metabolic pathways employed in kombucha-type fermentation, multiple aromatic compounds are referred to as microbial. However, although the beverage's final flavor maintains common compounds that contributes to the kombucha-type fermentation aromatic signature, typical aromatic compounds from its original substrate allow for the obtention of a fresh drink with innovative flavors. The Hakko Sobacha final flavor also depends on fermentation parameters such as the initial sucrose concentration and the liquid starter age and concentration, the microbial composition of the SCOBY, temperature, and time.

Study of Sensory Analysis Trends Resulting from the Hakko Sobacha Tasting
Initially, the consumer sample surveyed was predominantly male (57%), mainly in the 18-25 age bracket (54%) (followed by the 26-35 age bracket with 35%), and mainly university students (58%, versus 42% employees). In terms of consumption habits, 88% of consumers were familiar with fermented beverages. In contrast, 37% of the population had never tasted these drinks, followed by 28% who consumed them once a trimester and 18% several times a month. Kombucha was the most widely consumed beverage (87%), followed by kefir (45%). Finally, 75% of the population sampled bought these drinks in stores, and 55% made them at home.
The data resulting from the sensory analysis carried out on the formulated beverages were studied using PCA, as referred in Figure 7. However, since PCA explained only 36.2% of the total dataset variability in the 1:2 dimensions, a complementary analysis resulting from the triangular and hedonic tests was envisaged in order to draw more robust conclusions. The PCA variable dispersions regarding both drinks tested may illustrate the consumer's ability to perceive them differently. These trends were confirmed by the triangle test results, traducing the population's ability to differentiate the two drinks offered in the test (p < 0.0001). Descriptors concerning the drink with a kasha concentration of 10 g/L seemed to be grouped together in quadrant III, while those describing the 50 g/L kasha concentration were in quadrant II ( Figure 7A). In addition, the variable contributions explaining the data variability ( Figure 7B) might be the sourness and taste, followed by the roasted hazelnut aroma taste and the freshness of the drinks. The Spearman's correlation between the descriptors and kasha concentration modalities explained the descriptor contribution, as the quantitative score assessed by each judge regarding all descriptors allowed for the identification of links from one perceived descriptor to another ( Figure 7C). Considering the positive correlation between the sensorial descriptors, the most linked attributes were acidity intensity-sourness, then taste-roasted hazelnut aroma, and tastesourness. These results underlined the fact that the descriptors were correlated for a given beverage and were therefore perceived in different ways from one beverage to another. In this way, negative correlations therefore concerned descriptors for different beverages.
To go deeper into the beverage's evaluations and judge preferences, the individual clusters grouped by appreciation similarities as a function of biplot quadrants in the first two dimensions was envisaged ( Figure 7A). The QI cluster generally disliked the two drinks tasted for their acidity, taste, fizz, and hazelnut flavor. The QIII cluster, mirroring the previous group, seemed to like both drinks. A preference for the drink with the highest buckwheat concentration (50 g/L) was demonstrated by the intensity of the following variables: acidity and intensity, freshness, fizz, taste, and roasted hazelnut aroma. These outlines were confirmed by the pairwise test results, as the population proportion that preferred the highest buckwheat-concentrated drink for the following descriptors: acidity (63.33% (p < 0.001)), intensity (53.33% (p < 0.01)), fizz (55% (p < 0.01)), taste (63.33% (p < 0.001)) and roasted hazelnut aroma (73.33% (p < 0.0001)). From a global point of view, this beverage preference reached 57% (p < 0.001). The two remaining clusters seemed also opposed, where the group of individuals in the QII cluster appreciated both drinks for the characteristic's acidity, hazelnut aroma, intensity of acidity, taste, and freshness of drink_50 (kasha 50 g/L) on the one hand and taste, acidity, and hazelnut aroma of drink_10 (kasha 10 g/L) on the other. Individuals in the QIV cluster found the color of drink_50 pleasant but the acidity and hazelnut flavor unsatisfactory. Finally, the QIV cluster did not like drink_10 for the following characteristics: acidity, taste, intensity of acidity, hazelnut flavor, freshness, and fizz. The color attribute describing drink_10 (kasha 10 g/L) reached 53.33% (p < 0.01). There was no significant difference (p > 0.05) observed regarding the freshness descriptor (accounted for 51.67% and 48.33% concerning drink_50 (kasha 50 g/L) and drink_10 (kasha 10 g/L), respectively). Finally, the hedonic evaluation mentioned that the preferred times for Hakko Sobacha consumption were during the afternoon (67%) and as an aperitif (52%). The preferred quantity (71%) was the equivalent of a water glass (25 cL). The judges most frequent comments on the drinks were pleasant (44%), followed by a non-neglectable high Finally, the hedonic evaluation mentioned that the preferred times for Hakko Sobacha consumption were during the afternoon (67%) and as an aperitif (52%). The preferred quantity (71%) was the equivalent of a water glass (25 cL). The judges' most frequent comments on the drinks were pleasant (44%), followed by a non-neglectable high acidity and odor (15%). As a result, the major judges surveyed (80%) were ready to incorporate Hakko Sobacha into their daily routine as a more natural substitute for soft drinks.

Materials and Chemical Reagents
Kasha seeds used to prepare Sobacha infusion were sourced from Japan (Shizuoka). The Chun Mee Moon Palace green tea and Lien Son black tea leaves used in the fermentation liquid starter (fermented liquid obtained from a previous kombucha batch) were sourced from China (Zheijiang) and Vietnam (Yen Bai), respectively, and were certified to be organically grown (Da Zhang Shan cooperative, limiting the probability of undesirable molecules in the beverage). The sucrose used in fermentation was a refined white beet sugar from Tirlemont (Belgium). This prevented the formation of an off-flavor and boosted the alcohol content by stimulating the yeast activity throughout the fermentation process [7]. The growing medium used to ferment the tea was a symbiotic culture of bacteria and yeast (SCOBY) from Fairment (Berlin, Germany). This biofilm of cellulose was previously preserved in kombucha. Regarding the major fermentation component analysis, high-purity standards used for the external calibration (namely sucrose, glucose, fructose, ethanol, acetic and lactic acids) and the mobile phase solvent were obtained from Sigma-Aldrich (Darmstadt, Germany). The solvents used in this study were of analytical quality.

Hakko Sobacha Preparation
Sobacha solutions were prepared by infusing kasha at 10 g/L and 50 g/L in mineral water previously heated to 92 • C (Buffalo DM868; LACOR, Bergara, Gipuzkoa, Spain) [37]. After infusing for 20 min, the Sobacha solutions, made in three repetitions, were filtered and sweetened with white beet sugar (60 g/L) [2,37]. Once the solutions were cooled (<35 • C), the inoculum composed of a SCOBY and its liquid starter was added at concentrations of 1% and 10% (w/w), respectively. The liquid starter was the result of a previous kombucha fermentation [23]. The containers were covered with a gauze held by an elastic band (aerobic conditions) and then placed in an oven (IPP55plus; F-Nr.: V221.0308; Memmert GmbH+Co.KG, Schwabach, Germany) for the first fermentation (F1) for 10 days at 25 • C [19]. Then, solutions were filtered (200 mesh), and the second fermentation (F2) was set up by transferring them in mechanically sealed bottles (anaerobic conditions leading to natural carbonation) for a further 15 days at 25 • C. Measurements of kinetics were taken on days 0, 1, 2, 3, 6, 7, 8, 9, 10 and 25 of fermentation (referred to as D0, D1, D2, D3, D6, D7, D8, D9, D10 and D25 later in the manuscript). On each day of measurement, a volume of 10 mL was sampled and placed at −80 • C to carry out the different measurements and to evaluate the fermentation kinetics (Sections 3.3 and 3.4). At the end of F2, Hakko Sobacha beverages were kept in a cold chamber for sensory analysis (Section 3.5).

Dynamic Analysis of the Physico-Chemical Parameters of the Drink
The pH was measured with a pH meter (Hach BeRight; SensION+ pH1; S/N 605069; IP67, Hach Company, Loveland, CO, USA). The pH during fermentation was monitored via direct measurement in the fermentation medium on every sampling day.
At the end of F1, wet weights of nascent SCOBY pellicles were taken after 10 days of incubation. To do so, each nascent pellicle was separated from the original inoculum with sterile forceps, blotted on sterile paper towels over the pellicle 4 times and gently pressed to remove surface water [38]. Nascent pellicles were then dried until dehydration for 4 h at 65 • C (IPP55plus; F-Nr.: V221.0308; Memmert GmbH+Co.KG, Schwabach, Germany). Dry matter was then weighed. The analytical balance used was a Sartorius BC-EB (Sartorius Lab Instruments, Goettingen, Germany).
Quantification of the major components, namely carbohydrates (glucose, fructose, sucrose), ethanol (EtOH), and organic acids (acetic and lactic acids), was performed by high-performance liquid chromatography (HPLC). After sample thawing, centrifugation (Eppendorf MiniSpin, Eppendorf AG, Hamburg, Germany) was conducted with a volume of 2 mL at 10,000 rpm for 10 min at 25 • C [39]. The obtained supernatants were filtered with nylon syringe filters with a pore size of 0.22 µm. Assays were performed on an HPLC system manufactured by Agilent, equipped with an automatic liquid sampler (G1329A) with a thermostat temperature control set at 4 • C. The quaternary pump (G1311A) was used with a flow of 0.6 mL/min. For carbohydrate quantification, an Aminex HPX-87H column with dimensions of 300 mm × 7.8 mm (Bio-Rad Laboratories, Hercules, CA, USA) was used for the separation, with ultrapure H 2 O as the eluent. An Aminex HPX-87C column with dimensions of 300 mm × 7.8 mm (Bio-Rad Laboratories, Hercules, CA, USA) was employed for the separation of ethanol and organic acid kinetics, with 5 mM H 2 SO 4 as the eluant. Both columns were equipped with a Micro-Guard Carbo-C Cartridge with dimensions of 30 mm × 4.6 mm (Bio-Rad Laboratories, Hercules, CA, USA). Separations were conducted in isocratic mode, and the thermostatic column compartment (G1316B) was set at 50 • C and 85 • C for Aminex HPX-87H and HPX-87C, respectively. The sample injection volume was 10 µL and the analysis duration lasted 24 min. The detectors used were an MWD (multiple wavelength detector G1365D) set up at 210 nm, followed by an RID (refractive index detector G1362A) fixed at 30 • C. The components were quantified by external standardization.
These assays were performed on all the collected samples in order to obtain the kinetics (production/conversion) of these molecules of interest. Ethanol determination was controlled at the end of the fermentation process to ensure that the levels were below 1.2%, which is the European legislation level (1169/2011) for being considered as a non-alcoholic drink [12].

Chromatographic Analysis of the Development of Volatile Organic Compounds
VOCs were characterized using the SBSE-GC-MS technique, as described by Suffys et al. (2023) with minor modifications [23]. The column used for the separation was a capillary non-polar HP-5MS (5% Phenyl Methyl Polysiloxane) with dimensions 30 m × 250 µm × 0.25 µm.
To optimize compound identification and data processing, the following filters were applied as a first rough compound sorting of the raw data: match factor (percentage of probability of correct identification of the compound under consideration) ≥75% and maximum area under the curve ≥5%. Next, a manually disciplined treatment approach consisted of the validation of all molecules one by one, referring to the mass spectrum by consulting the spectra of reference molecules in the literature. Additionally, the retention index was finally considered to match the validation hypothesis [40].
The semi-quantification of VOCs was calculated relative to the injected internal standard. The determination of OAV (odor-activity value) was calculated as a function of the considered compound concentration in the sample divided by its ortho-nasal detection perception threshold in water, as referred to in the literature [41,42]. This concept allowed for the study of the contribution of the compounds to the aroma of the drink [23,43].

Sensory Analysis
The sensory analyses carried out in this study were the following: triangle discriminative test and appreciation test (hedonic). A questionnaire comprising the initial personal information about the judge, both triangle and hedonic tests, and then questions analyzing his or her knowledge of fermented beverages and consumption habits, ending with questions about preferences for integrating Hakko Sobacha into daily life being proposed. The survey was conducted with a representative sample of 60 evaluators. For this purpose, a recruitment invitation was sent to staff members and students of the Gembloux Agro-Bio Tech faculty (University of Liege, Belgium). The samples tested were the final formulated drinks, corresponding to the last fermentation sampling day (D25), for both buckwheat concentrations considered (10 g/L and 50 g/L). Both beverages (4 cL at 4 • C) were served with water and crackers (mouth rinsing) to allow the untrained jury to taste a panel of products without experiencing any sign of saturation. For representative sampling of the product, a blind presentation of the samples was carried out using a neutral container, with minimum information given about the beverages, and coding with a random 3-digit number. Samples were presented in a homogeneous way and randomized order [44,45].

Triangle Sensory Analysis
The discriminatory triangular test was used to test the hypothesis of identity between two fermented drinks, characterized by their different starting kasha concentrations in the infusion (10 and 50 g/L). The three samples (which had two identical ones) were served (4 cL, 4 • C) simultaneously in a randomized and balanced form [46][47][48][49]. The judge therefore had to determine which of the two samples was different (one of which was presented in duplicate). Values of β = 0.05% and p d = 0.20 were chosen to ensure that no more than 20% of the population could detect a difference among the samples (within a 95% confidence level). As the results obtained referred to the evaluators' ability to perceive the difference or otherwise between the two drinks, the data were analyzed by the normal approximation of the binomial test to characterize the judges' behavior when faced with this test. The critical value table for the triangle test was used to assess the presence of any significant differences [48,49].

Hedonic Evaluation
A non-specific hedonic test was considered in two parts. First, an acceptability measurement was used to evaluate the appreciation of the products on a scale of criteria [50]. Color, taste, acidity, sparkle, roasted hazelnut aroma, and freshness descriptors were evaluated [39,51]. For this purpose, a 5-point appreciation scale (from 1, the lowest preference, to 5, the highest preference) was used. Secondly, a preference measurement was envisaged by a pairwise test, allowing the judge to assign a preferred drink for each descriptor studied. The presentation of the samples (4 cL at 4 • C) was carried out monadically and then simultaneously, respectively.

Raw Data Analysis
All chemical analyses and fermentation medium measurements were performed in three replicates, and the values were expressed as means (n = 3) ± standard deviation (SD). Basic statistical data processing was performed using GraphPad Prism (8.0.1 software, GraphPad software Corporation, San Diego, CA, USA) and Microsoft Office Excel (Excel 2021 software, Microsoft Corporation, Redmond, Washington, DC, USA). Multiple student's t-tests were performed on the pH data, the SCOBY weights and the HPLC data to determine if there were significant differences (n = 3; α = 5%) measured between both the buckwheat concentrations and the stages of fermentation modalities. Moreover, this statistical test type was used to identify significant differences (n = 60; α = 5%) regarding the hedonic evaluation and pairwise test consumer behaviors.
Principal component analysis (PCA) was performed on all VOC concentrations, considered as individuals. Hakko Sobacha fermentation stages and kasha concentrations were represented as vectors. Considering the sensory analysis, PCAs were performed with evaluators as individuals and samples descriptors as variables. PCA using the Spearman correlation matrix was conducted on the Hakko Sobacha sensory descriptors and kasha concentrations as variables.
The principal components PC1 (Dim1) and PC2 (Dim2) were considered and reported in this article. The variable representation quality was illustrated as cos 2 values, rated by contribution scale. PCAs were performed in R (R 4.0.2 software, R Development Core Team, Boston, MA, USA).

Conclusions
To conclude about this survey on the development of a new functional drink based on Sobacha infusion, the main kinetics including pH, SCOBY yields, major fermentation components, VOCs, and aroma-active compounds were analyzed during Hakko Sobacha fermentation. Moreover, the sensory analysis of the formulated beverages reflected the consumer preferences and acceptance of these new refreshing drinks.
Concerning the physico-chemical characterization of kombucha, the pH decrease traduced the acidification of both fermentation media and thus the microorganism's ability to metabolize diverse products in terms of Sobacha composition. Moreover, the low pH conditions ensured food safety by inhibiting any pathogen development, considering both formulated drinks.
The evolution of the SCOBY weights as a function of the buckwheat concentration pointed out a correlation between the different acidification profiles obtained from the kasha concentration modalities and the carbon sources available for cellulose production during the aerobic fermentation step. However, a study focusing on the comparison of SCOBY production for Sobacha, and tea processing could reveal the performance of the fermentation system under consideration.
The major compounds' kinetics were analyzed using HPLC, and the main microbial consortium routes were highlighted, such as the glycolysis pathway and AAB pathways. The buckwheat concentration therefore seemed to have an impact on the metabolic activity of sugar conversion by the microorganisms during the first fermentation. Since Sobacha is a food matrix that is rich in starch and made up of glucose units, the main source of fructose in the fermentation medium is sucrose. The availability of additional glucose from the starch hydrolysis during the anaerobic step allows for obtaining a higher final ethanol content-both beverage's ethanol levels remaining below the 1.2% (v/v) considered by the European regulation (1169/2011) for characterizing non-alcoholic drinks.
Dynamic changes in VOCs were investigated using SBSE-GC-MS during Hakko Sobacha fermentation. The fermentation media were mainly composed of alcohols, aldehydes, carboxylic acids, esters, ethers, ketones, phenols, pyrazines, and terpenes. Pyrazines present in the infusion were for the most part related to the buckwheat seed roasting process. Thus, their levels were positively correlated with kasha concentration.
Furthermore, the highest concentrated infusion tested seemed to enrich the diversity of the molecules developed during fermentation. The fermentation process dynamically changed the VOC Sobacha profile into a different mixture of compounds. The synthesis of 2-phenylethyl acetate and ethyl caprylate participated in the kombucha-type fermentation signature. The major phenol found in Hakko Sobacha is 4-ethylguaiacol, an aromatic phenol mainly produced by the yeast genus Brettanomyces.
The aromatic analysis, based on odor-active values, accounted for about 30 aromaactive compounds. Sobacha presented fat-citrus-green, creamy-waxy, sweet-fruit, and woody-spicy-cumin-oregano aromas. Moreover, typical kasha aromas are correlated with the infusion concentration, bringing flavor such as peanut butter-wood-nut and cocoa. During the fermentation process, the Sobacha sensory profile was modified through MO activity, resulting in the development of diverse aromatic molecules with different perception profiles. The synthesis of 2-phenylethyl acetate, ethyl caprylate and pelargonic acid contributed to the final beverages' flavors with a pineapple aroma, sweet-waxy-fruitypineapple notes, and waxy-dairy-cheesy aromas, respectively.
Concerning the sensory analysis results, the PCA variable dispersions regarding both drinks tested seemed to illustrate the consumer's ability to perceive them differently, as confirmed by the triangle test results. In addition, the descriptor contributions explaining the data variability were the sourness and taste, followed by the hazelnut aroma taste and the freshness of the drinks. Considering the positive correlation between sensorial descriptors, the most linked parameters were acidity intensity-sourness, then taste-roasted hazelnut aroma, and taste-sourness. Moreover, the hedonic evaluation mentioned that the judges' most frequent comments on the drinks was pleasant. Broadly speaking, as both beverages are appreciated by consumers for their respective descriptor scores, no beverage was rejected from this study. As a result, the major judges surveyed were ready to incorporate Hakko Sobacha into their daily routine as a more natural substitute for soft drinks.
From an industrial perspective, as the fermentation process is affected by many parameters such as temperature, pH, oxygen level, dissolved CO 2 , system precursor supply, shear rate in the fermenter, and the nature and composition of the medium, variations in these factors could affect fermentation speed, performance, organoleptic properties, nutritional quality, and other physico-chemical properties of the product.
Besides all the parameters mentioned above, the kombucha-type fermentation production process might also affect its final properties. Until now, the process has still been rather traditional, and the exact proportion of components may differ according to the product desired.
Nevertheless, to optimize the industrial production of Hakko Sobacha as a functional beverage, a comprehensive study including large-scale production, microbiological identification, and biological testing should be carried out.
As far as prospects are concerned, the characterization of the microbial beverage population could confirm some existing relations between VOC synthesis and the metabolic pathways involved during fermentation, not only by identifying the microorganisms present in the starting SCOBY but also by studying its evolution and potential specification regarding the matrix adaptation.
The SCOBY morphology is also an indicator of the variability of the microbial community. The appearance of the SCOBY could also be studied in greater depth by analyzing its cellulose composition, the formation of sheets, intermolecular links leading to the synthesis of another biopolymer, and the potential capacity for branching cellulose.
As far as the health aspects of the beverage are concerned, it would be interesting to investigate the potential antioxidant and anti-inflammatory properties of Sobacha. Indeed, the evolution of these parameters with fermentation modalities could help to develop optimal functional beverages. Moreover, interactions between the beverage and the intestinal consumer microbiomes should be further studied.
A panel of functional beverages of this type could help in determining the limit of the necessary nutrients for SCOBY development and the dependance factors regarding diverse fermentation parameters. The study of matrix constituent kinetics may allow for the evaluation of this fermentation substrate's impact on health.
Finally, the economic aspect could be considered in terms of commercializing Hakko Sobacha, as the initial kasha matrix is quite costly. On the one hand, a reduced kasha content could be set. Secondly, as importation from Japan represents a large part of the matrix costs, the development of a local kasha production chain could be envisaged. This would involve studying the parameters for preparing buckwheat and setting up a process for grain roasting, using Japanese know-how to ensure that the matrix is rich in flavor and has beneficial health properties, not to mention its acceptability in non-Asian countries.

Institutional Review Board Statement:
The study was conducted in accordance with the Declaration of Helsinki, and panelists gave informed consent before participating in this study.
Informed Consent Statement: Informed consent was obtained from all subjects involved in this study.

Data Availability Statement:
The datasets generated for this study are available on request to the corresponding author.