Virtually Together: Exploring the Impact of a Low-Latency Synchronized Audio Platform on Participant Wellbeing in a Facilitated Online Singing Group

Abstract

While previous research shows that group singing has clear benefits for wellbeing, there is limited data on how these benefits transfer to online singing groups. A common challenge in virtual settings is that high-latency audio delays are incompatible with collective music making. Therefore, many online singing groups use a format in which the facilitator is heard collectively, while participants are muted. Emergent technologies now offer low-latency audio options that enable group-singing participants to hear each other synchronously. In this mixed-methods study, we compared the physiological and emotional effects of participation in a facilitated online singing group comprised of participants with neurological conditions, care partners, and other older adults. We used both modalities in succession: high-latency muted audio and low-latency synchronous audio. We investigated how being heard and hearing each other modulates participant experience and wellbeing. We measured changes in participants’ mood, reported pain, and feelings of social connection via questionnaires administered before and after each session. Salivary samples were collected before and after selected sessions to monitor cortisol levels. Additionally, focus groups and individual interviews provided qualitative insights into participants’ experiences. Results indicated that online group singing in either format led to increases in positive mood and social connection. Notably, the synchronous audio sessions resulted in greater enhancements in social connection and a trend towards larger improvements in mood. Singing was associated with decreased self-reported pain and lower salivary cortisol, with no significant difference between the two platforms. Qualitative findings revealed that once technological barriers were addressed, hearing each other synchronously significantly enhanced social connection and musical enjoyment. Concomitantly, results also revealed benefits unique to the muted audio modality. This study contributes evidence that virtual group singing offers a viable format for those whose circumstances preclude in-person activities and highlights a spectrum of meaningful collective singing and social connection experiences.

Introduction

Group singing has been shown to have clear benefits for mental and physical wellbeing for people across the lifespan (Camlin et al., 2020; Johnson et al., 2013; Maury et al., 2022; Prendergast et al., 2009), and has been increasingly used in a therapeutic context to enhance quality of life and improve symptoms in disorders ranging from mental health conditions (Williams et al., 2018) and dementia (Baker et al., 2022; McDowell et al., 2023; Osman et al., 2016) to Parkinson's Disease (Good et al., 2023; Mallik et al., 2025; Tamplin et al., 2024) and respiratory conditions (Smallwood et al., 2022; Philip et al., 2024).

Virtual Group Singing

The onset of the COVID-19 pandemic in early 2020 led to widespread social lockdown, abruptly limiting or halting in-person community arts activities, including group singing, which many relied on for social connection. The loss of these communal music experiences was profoundly felt during a time of increased isolation.

Into that breach, the expansion of virtual group platforms provided new opportunities for social and creative connections worldwide (Chen et al., 2022; Tamplin & Thompson, 2023), generating innovative artistic collaborations and multi-modal performance creations. Transitioning to virtual platforms posed an unexpected challenge to musical practices normative to in-person singing collectives such as choirs or facilitated therapeutic group-singing activities. The high-latency audio that is standard for group social platforms generates a slight sound reception lag that is suitable for conversation modalities, but which precludes the audio synchrony necessary for voices to align in pitch unity – a fundamental element of many collective singing practices. Such latency challenges disrupt both group-generated harmonies and overall musical flow. The resultant dissonances are incompatible with the nature of the activities, and, in the therapeutic realm, might also instill a sense of discomfort for some participants. We conceptualize latency as influencing the capacity for musical synchrony, which in turn shapes social connection and subjective wellbeing.

The effects of latency on networked music performance have been studied for the last several decades, and it is well-recognized that latencies over 50 msec tend to disrupt the ability of virtual performers to synchronize. For instance, Chew et al. (2004) reported on a piano duo playing Poulenc's Sonata for Piano Four Hands over a networked connection, and found that delays of less than 50 msec were tolerable, but longer delays became increasingly difficult. Chafe et al. (2010) found that delays of ∼25–60 msec led to a deceleration of tempo, and at delays of >60 msec, the performance tended to fall apart. Other studies of pairs of musicians clapping or playing music together over networked platforms led to similar conclusions (Bartlette et al., 2006; Driessen et al., 2011; Rottondi et al., 2015). With practice, experienced performers have been able to find strategies to adapt to slightly longer delays (Chew et al., 2005; Cheston et al., 2024). The length of the delays caused by typical videoconferencing platforms (∼200 msec, as reported by Cheston et al., 2024) is far too long to allow any such adaptation, making real-time virtual choral activities difficult.

During the COVID-19 pandemic, rather than abandon their group activities, singing collectives adapted by having choir conductors lead muted participants in online rehearsals. A related format was to create video compilations using pre-recorded individual contributions rather than real-time performances. Notably, inspiring mass choir collaborations emerged on platforms like YouTube and were enjoyed by millions (e.g., Whitacre, 2020).

While participants could not hear each other singing in real time, online group singing using muted audio still proved beneficial in reducing loneliness and stress, and in enhancing mood and life satisfaction (Schäfer, 2023). Studies on the effects of virtual music making are relatively limited at the time of writing. Grebosz-Haring et al. studied a group of healthy adolescents and compared in-person singing sessions which occurred in early 2020 to virtual sessions that occurred later in 2020. Results indicated that both types of session had positive impact on mood, but that in-person sessions led to decreases in cortisol and greater reductions in reported stress levels (Grebosz-Haring et al., 2022). Similarly, a study published in 2021 surveyed participants in musical groups that had pivoted to online sessions during the pandemic and found that virtual music making led to significantly lower scores for group identification and psychological needs satisfaction (Draper & Dingle, 2021). A survey of health-focused choir facilitators about how they adapted during the COVID-19 pandemic reported that the majority of choirs moved to an online format, with the expected negatives including the lack of ability to hear each other sing, and frustrations with technology. However, there were unexpected benefits such as the possibility for longer sessions due to the lack of travel time. Some participants preferred being able to sing without the pressure of being heard by others (Tamplin & Thompson, 2023). Overall, the consensus from previous studies is that while there are some advantages to muted audio online singing sessions compared to in-person, the psychological effects of muted virtual singing may be weaker than those of in-person singing.

As COVID-19 distancing restrictions were suspended, music groups quickly resumed in-person gatherings, thereby rendering the option of online audio technology seemingly moot for most. However, a consummately important outcome of the technology has concurrently become clear: the realization that for a large subsector of group-singing enthusiasts – those with medical, mobility, or other access challenges, the virtual realm now offers perpetually available resources for remote music making. We have in fact seen the continuation of therapeutic online singing groups for people with Parkinson's disease, COPD, post-natal depression, and dementia (Bind et al., 2023; MacDonald & Zumbansen, 2023; Philip et al., 2024; Tamplin et al., 2024).

Synchronous (Low-Latency) Audio

The achievement of virtual low-latency audio technology, through which the time lag was sufficiently eliminated to enable synchronous music making, had been reported since the previous decade (Cáceres & Chafe, 2010; Timson, 2021). Explorations by professional and student musicians were undertaken throughout the COVID-19 era (e.g., Bosi et al., 2021). The dynamic expansion of public access to low-latency virtual audio technologies has opened options through which virtual singing collectives can hear each other synchronously in real time, providing game-changing venues for musical connection in virtual meeting spaces.

However, the literature regarding the use of synchronous audio technology platforms for group singing collectives remains limited, and the purposeful implementation possibilities remain relatively unknown and unsourced among facilitators of group-singing collectives. At the time of writing, the majority of these groups continue to use a muted format on high-latency videoconferencing platforms. With increased awareness of accessible low-latency synchronous audio technology platforms, those whose circumstances prevent in-person activities can encounter the intrinsic health and wellbeing benefits of collective singing, and can experience artistic co-creation through expression and connection.

The pandemic led to popularization of these tools for studio musicians and performers but there has been little reported use of the technology in choirs. To our knowledge, there has only been one published paper examining the use of low-latency technology in a choir and this reported on feasibility and proof-of-concept rather than on benefits and participant experience (Chen et al., 2022). An unpublished study (Pachete, Good, & Russo, in preparation) showed that group-singing sessions using low-latency technology provided indications that low-latency group singing offered a level of social connectedness and related wellbeing that was commensurate with in-person group singing.

Research Design

This research article presents the results of a proof-of-concept pilot study using low-latency virtual synchronous audio in facilitated remote-access group-singing activities within a health and wellness arts program. The study was designed to investigate the implementation process, practical production, and logistical challenges, which are not reported in full detail here. The overall experience and emerging benefits for the participants are the focus of this paper. The quantitative data and qualitative participant narratives provide valuable information for group-singing facilitation, and for a community of practitioners and researchers across the arts, sciences, and humanities continuum. Our group is a hub of SingWell, the overarching research network which supports this study. SingWell fosters multi-disciplinary research collaborations focused on group singing in medical and wellness contexts. Our team included experts from ethnomusicology, neurology, neuroscience, nursing, medicine, psychology, and various music disciplines, such as education, performance, pedagogy, and care-based singing facilitation. This collaborative fusion of knowledge enriched our understanding and sparked new avenues for inquiry and innovation.

The current mixed-methods study followed a 12-week singing class as it transitioned from a high-latency audio format which required participants to mute themselves, to a low-latency audio format where participants could hear each other sing synchronously. The concept for this research arose from a pre-existing “musical engagement” class that launched virtually on the Zoom platform in 2020, within the BC Brain Wellness Program, situated at the Djavad Mowafaghian Centre for Brain Health at the University of British Columbia. The BC Brain Wellness Program combines clinical care with arts-based interventions that support healthy lifestyles, striving to improve and sustain quality of life for people living with chronic brain conditions as well as that of their care partners, while also empowering healthy agers to enhance wellbeing and engage in preventative lifestyle measures. The diverse multi-strand nature of this class community as a recruitment field for the study’s participant group offered an unusual opportunity to explore dynamic aspects of music as a care tool in virtual spaces.

Rather than comparing two separate singing groups, each implementing only one of the audio technologies, we opted to examine a comparison within one singing collective as it experienced both modalities in succession. Beginning with the muted audio modality on a high-latency platform, the participants transited midway through the course to explore the unmuted synchronous group-singing modality on a low-latency platform.

The design format enabled us to gather rich qualitative data about the participants’ social and artistic experiences within both individual platforms, as well as the transition process between the two. Relatedly, the post-study focus group and interview data provided powerful testimonials regarding the moment when participants shifted into shared musical space, and data regarding their perceptions of the contrasting and similar elements of practice inherent within the two modalities. To our knowledge, this is the first study to directly compare muted versus synchronous low-latency online facilitated group singing using mixed methods, integrating biological, psychological, and social outcomes.

Research Questions and Hypotheses

The nature of this study encompassed a range of questions relating to technology implementation, adaptations to in-situ facilitation and pedagogical practices, and the experiential shifts of the participant collective. We investigated the following: 1) Does online group singing using a synchronous audio platform have expanded benefits to wellbeing as measured quantitatively by reported mood, pain, and feelings of social inclusion, as well as stress biomarkers? 2) How did the shared musical experience influence the participants’ sense of connection as measured through qualitative interviews and focus groups? Overall, we wanted to understand how the capacity to hear and be heard synchronously and collectively alters the nature and outcomes of online facilitated group singing. That is, how does the introduction of low-latency audio technology change quantitative measures, and what is the experiential impact for participants through these diverse online group-singing sessions upon their social, emotional, and physical wellbeing?

Based on existing studies which demonstrated that group singing has benefits for mood and stress in both in-person (Bullack et al., 2018; Gick, 2011; Good & Russo, 2022) and online (Schäfer, 2023) formats, we anticipated that group singing on either platform would lead to improvements in measured aspects of wellbeing (increased positive mood, decreased pain and stress). We hypothesized that the introduction of low-latency audio synchrony would increase the sense of connectedness when compared to muted high-latency audio sessions in which participants were able to hear only the facilitator's voice in tandem with their own. A number of previous studies have shown that group singing facilitates social connection (Good & Russo, 2022; Pearce et al., 2015; Lonsdale & Day, 2021; Weinstein et al., 2016). Pearce et al. (2015) studied participants in adult education classes and found that all types of classes studied saw an increase in closeness over time, but that singing classes increased in closeness ratings much faster than the other groups. In muted audio online singing groups, participants can see one another, but do not experience the same social benefits as when singing in-person (Draper & Dingle, 2021). We hypothesized that it is the capacity to hear one another singing that allows for shared collective musical expression, in which individual singing voices coalesce into synchronies and foster deeper musical connection. Conversely, we also projected that the transition might present mixed outcomes. For example, some participants might feel uncomfortable being heard collectively and thus experience a sense of relief in staying muted.

For many, singing collectively is among the most common and natural forms of human connection and may contribute powerfully to a sense of wellbeing. For those with medical or other access challenges, the ability to participate in a singing group is expanded by the availability of technologies that allow online singing. In this work, we have explored the provision of a vibrant, interactive musical activity via two different online audio modalities and have chronicled the participants’ perspectives both qualitatively and quantitatively. The information gathered during our exploratory pilot may contribute beneficial tools for the implementation of facilitated collective singing experiences in both audio modalities within the virtual space. The low-latency synchronous audio modality offers many directions of opportunity to increase musical connection and wellbeing. It is our hope that this study's experiential data and outcomes contribute to the cultivation of the virtual group-singing community within the music-care continuum.

Methods

All recruitment and study procedures were conducted with approval of the University of British Columbia Behavioural Research Ethics Board.

Participants

Eighteen participants were recruited through community partners and social media posts. Eligible participants included adults with a chronic neurological condition, care partners for a person with a chronic neurological condition, and persons from the general community who were at least 50 years of age. Participants were required to live in geographical proximity to the study center, be fluent in English, and have access to a computer or tablet able to run current versions of the required software. Participants were excluded if they had conditions that might cause oral bleeding, as this could interfere with salivary cortisol analysis.

Class and Class Formats

From September to December, 2023, subjects participated in 12 one-hour online singing sessions, facilitated by an experienced music educator. Class components included guided deep breathing exercises, intentional listening, postural alignment, vocal warm-ups, hand symbols representing vowel sounds, movement responses to rhythm, collaborative shaping of phrases, singing of familiar songs, sharing of personal stories inspired by songs, and learning new repertoire. The first seven sessions were conducted using Zoom, a high-latency teleconferencing platform. To address the issues inherent to audio synchrony lag (∼200 msec, Cheston et al., 2024), participants used the mute function, allowing them to hear the facilitator singing and playing the piano but not the other singers. Participants unmuted themselves between songs for discussion and social interaction and were able to see the facilitator and other participants throughout the session. The subsequent five sessions were conducted using JackTrip synchronous audio-visual software (Cáceres & Chafe, 2010), a low-latency platform which allows participants to be heard synchronously with minimal delay (< 40 msec) between voices. As our study commenced, the technology was in the midst of a major upgrade transition from a necessary external bridge device to a software-only application system which was only available for Mac systems. Half the participants required the use of a bridge device, while the other half used the native Mac application. All were provided USB microphones and headphones as needed. All necessary software/hardware was delivered and installed in participants’ homes by research assistants starting in week 4. Both the external bridge and Mac application formats required stable, high-speed wired internet access and led to similar latency times (<50 msec). Installation delays—caused by ISP blocks and emergent software issues with bridge devices—necessitated the group's continued use of the high-latency platform for a seventh session, instead of the planned six sessions per modality.

Quantitative Questionnaires

Questionnaires were administered online using a Qualtrics platform. Participants completed a demographic and baseline information survey prior to the first session. Before and after every session, they completed a Likert-scale based survey to assess mood via selected emotions from the Positive and Negative Affect Schedule (PANAS, in which people rate how well different adjectives describe their mood on a 5-point scale from “Not at all” to “Extremely”; Watson et al., 1988), pain (7-point scale from “not pain” to “extreme pain”), connection to the group (7-point scale from “not connected at all” to “very connected”), connection to the world (7-point scale from “not connected at all” to “very connected”), and the Inclusion of Other in Self scale (IOS, a 7-point visual scale; Aron et al., 1992). Positive affect was calculated as the composite score of the participants’ responses to questions about how much they felt excited, strong, enthusiastic, proud, inspired, and active. Negative affect was calculated as the composite score of the participants’ responses to questions about how much they felt irritable, nervous, and jittery. After each session, participants were also asked to rate their overall experience of the session and the effectiveness of the session in creating a sense of connection. Due to a communications error, Likert scales were not collected immediately prior to the first session; session 1 is therefore not included in our quantitative results.

Salivary Cortisol

Participants were given a demonstration and written instructions on how to collect salivary samples using the passive drool method. Saliva was collected before and after sessions 4 and 6 of the muted audio condition and sessions 1, 3, and 5 of the synchronous audio condition. Participants collected saliva into cryotubes using a saliva collection aid (Salimetrics Inc.). Participants were instructed to avoid eating, smoking, or brushing their teeth for one hour before collection, and to rinse their mouth with water just prior to collection. They were instructed to immediately place the vials in their freezer. Within 5 weeks of collection, salivary samples were collected from participants’ homes and frozen at −20 °C until assayed. Cortisol concentration was assayed in-house using a Salivary Cortisol ELISA Kit (Salimetrics Inc.) following the manufacturer's protocol.

Statistical Analysis

Our statistical analysis of questionnaire data focused on sessions 2–5 of each condition. We did not have a complete set of data from session 1 of the muted audio condition due to a communications error; therefore session 1 was not included in the analysis. Sessions 6 and 7 of the muted audio condition were not included in the analysis because there were no corresponding synchronous sessions.

The outcome measures were: Positive Affect (PA), Negative Affect (NA), Inclusion of Other in the Self (IOS), Connection to Group, Perceived Pain, and Cortisol. Our objectives were to examine: 1) the overall effects of group singing (pre vs. post); 2) the effects of condition (muted audio vs. synchronous audio); and 3) the impact of number of group-singing sessions (Number of Sessions). For objective 1, mean differences between pre and post measurements were used. For objectives 2 and 3, change scores were used and calculated by subtracting the scores measured before each session (pre), from scores immediately following the session (post). Finally, before statistical modeling, outlier rejection was conducted for cortisol data given its continuous nature. Outliers were rejected based on a boxplot method, which rejects datapoints which fall between ±1.5 interquartile range. A total of 16 data points (11%) were flagged as outliers and rejected.

All statistical modeling was conducted using multilevel models using the lmer4 package (Bates et al., 2015) in RStudio (Posit Team, 2025). The variables of interest included Time, i.e., before or after singing (pre vs. post), Condition (synchronous audio vs. muted), and Number of Sessions in each condition (sessions 2–5). As each variable of interest was categorical, IVs were coded using contrast codes with one level used as a referent. Referents were set as: pre (Time), muted audio (Condition), and session 2 (Number of Sessions).

To determine the overall effect of group singing (Objective 1) we defined a model which looked at mean differences between pre and post scores (DV ∼ Time). To determine the effects of synchronized audio (Objective 2) and how these effects varied over multiple sessions (Objective 3), we defined a model which contained Condition, Number of Sessions, and an interaction term between the two (DV^change ∼ Condition + Number of Session + Condition*Number of Sessions). All main effects are reported and in the case of a significant interaction term, simple slope analysis is used to examine how differences between Condition varied across Session Number. For tests of significance, thresholds in alpha values were set at 0.05. For significant effects, effect sizes are reported using Cohen's d.

Focus Groups and Interviews

Voluntary focus groups were conducted via recorded Zoom sessions during weeks 6 and 7 of the study as well as after week 12. Additionally, one-on-one interviews were conducted post study. 11 participants took part in the first set of focus groups (FG1), 13 participants in the second set of focus groups (FG2) and 12 participants in the individual interviews (II). These qualitative dialogues were guided by the following conversational prompts and questions that related to the participants’ experience:

How would you describe the experience of singing without (with) hearing others throughout this first (second) half of our sessions together?

Foundations of an evolving curriculum of care & belonging lie in the pursuit of each musical encounter with creativity, curiosity and intuition, more so when in collaboration with others. Was there a moment during our sessions these past few weeks when you were encouraged by someone else's creative expression while singing? If yes, please provide more details of that experience.

Music has a unique power to connect us deeply and to enhance our sense of cohesion when we are musicking together. How connected did you feel to other participants during the singing sessions in this half of the study? Please provide more details of that experience.

Musical expression is a form of art and, as described by one UBC music education scholar, is an experience that is of “more than ordinary significance” (Goble, 2021). Has there been some aspect of our singing together that has offered deep meaning to you? If yes, please provide more details of that experience.

The researcher who conducted the focus group sessions read these prompts and questions verbatim and then elaborated so that all participants understood the questions.

Qualitative Analysis

Focus group and interview sessions were transcribed from Zoom recordings and were hand coded. For each transcript set, half of the transcripts were read, and a list of initial codes and concepts were generated (Braun & Clarke, 2006; Miles et al., 2020). Using the procedures of constant comparison analysis (Grove, 1988), descriptive codes were assigned to segments of text that reflected a salient aspect of the participants’ experiences. The codes were assembled into categories based on shared meanings and the codes and concepts were then used for the second half of the transcripts. After the initial coding of the transcripts, a reflective summary document was created which outlined the codes, concepts, and broad, emergent themes. Then each transcript was read a second and third time to further refine the categories and continue to identify higher level themes. Codes and categories were compared throughout the analysis and new codes added as needed. The categories were examined for interrelationships and further synthesizing into major themes that reflected the nature of the participants’ experiences (Thorne, 2000).

Results

Participants

Eighteen individuals participated in the study. Seven participants self-identified as having a neurological condition, three were care partners of persons with a neurological condition not involved in the study, and eight identified as persons over the age of 50 who neither had a neurological condition nor were a care partner. The average age of participants was 66.6 ± 10.1 (range 49–84). Further summary characteristics of our participant population are presented in Table 1. Three participants dropped out of the study at or before the half-way point, due to personal reasons.

Table 1.

Participant characteristics.

Age (years)		66.6 ± 10.1
Sex	Male	4 (22%)
	Female	14 (78%)
Type of participant	Neurological condition	7 (39%)
	Care partner	3 (17%)
	Age > 50 and not one of the above	8 (46%)
Level of education	High school	1 (6%)
	College	1 (6%)
	University undergraduate	6 (33%)
	University postgraduate	10 (55%)
Overall health (self-report)	Poor	1 (6%)
	Fair	5 (28%)
	Average	4 (22%)
	Good	6 (33%)
	Excellent	2 (11%)
Number of people in household	One	5 (28%)
	Two	8 (44%)
	Three	3 (17%)
	Four	1 (6%)
	Five or more	1 (6%)
Other regular singing activity	Yes	4 (22%)
	No	13 (72%)
	No response	1 (6%)
Do you sing or hum when alone?	Yes	16 (89%)
	No	2 (11%)

Numbers are reported as mean ± standard deviation or n (%).

Quantitative Results

Positive Affect (PA)

Participants reported significantly higher levels of positive affect after the singing activity in comparison to their pre-singing state (d = 0.45; Figure 1A). This indicates a modest but reliable improvement in positive affect following group singing, consistent across sessions and conditions. When examining whether these changes varied by session or singing condition, no significant interactions emerged (Figure 1B; Table 2). However, a positive trend that was approaching significance suggested greater increases in positive affect for the synchronous audio condition compared to the muted condition (p = .08), hinting that the synchronous audio condition may lead to greater benefits to positive affect, compared to muted audio. No differences were observed between sessions when pooled across conditions. Lastly, based on visual trends we conducted a post-hoc analysis to examine differences between conditions across sessions. This revealed a significant difference between synchronous audio and muted conditions for session 4 of each condition (d = 0.2; Figure 2), but not for any other sessions. Taken together, these findings indicate that group singing produces consistent gains in positive affect, with tentative evidence that synchronous audio may amplify this effect, but only after participants have gained experience with the software.

Figure 1.

Violin plots showing changes in positive affect. (A) Positive affect scores measured before (pre) singing sessions were significantly different from those measured after (post) singing sessions, p < 0.001. (B) Change in positive affect (post-pre), when collapsed over session, showed a difference between conditions that approached but did not reach significance. Points in the violin plots represent means, and error bars represent bootstrapped 95% confidence intervals. Both are generated by the ggplot function stat_summary, which by default uses 1000 repetitions to determine confidence intervals.

Figure 2.

Change in positive affect (post-pre), shown by session and condition. Positive affect change peaked in Session 4 of the Synchronous audio condition, and was significantly different from the change in positive affect in Session 4 of the Muted audio condition. Points in the violin plots represent means, and error bars represent bootstrapped 95% confidence intervals. Both are generated by the ggplot function stat_summary, which by default uses 1000 repetitions to determine confidence intervals.

Table 2.

Summary of results of multilevel models across outcome variables.

Outcome measure	Time (Pre vs. Post)	Condition (synch. Audio vs. Muted)	Interaction (Number of Sessions * Condition)
Positive Affect	b  = 3.4, t(191) = 7.01, p < 0.001***	b = 1.62, t(91) = 1.77, p = 0.08	F(80) = 0.6, p = 0.61
Negative Affect	b  = −1.1, t(191) = −5.88, p < 0.001***	b = −0.22, t(89) = −0.68, p = 0.49	F(83) = 0.2, p = 0.9
IOS	b  = 1.09, t(191) = 6.33, p < 0.001***	b = 0.2, t(88) = 1.2, p = 0.23	F(82) = 1.17, p = 0.32
Connection (Group)	b  = 0.78, t(190) = 5.57, p < 0.001***	b  = 0.48, t(88) = 2.79, p = 0.006**	F(82) = 1.6, p = 0.19
Perceived Pain	b  = −0.23, t(190) = −2.01, p = 0.04*	b = 0.2, t(88) = 0.14, p = 0.89	F(82) = 0.08, p = 0.97
Cortisol	b  = −0.02, t(82) = −1.95, p = 0.05*	b = −0.03, t(42) = −1.08, p = 0.28	F(36) = 0.29, p = 0.58

	Session
Outcome measure	Session 2 vs. Session 3	Session 2 vs. Session 4	Session 2 vs. Session 5
Positive Affect	b = 0.58, t(87) = 0.45, p = 0.65	b = 1.35, t(86) = 1.04, p = 0.3	b = −0.88, t(87) = −0.7, p = 0.49
Negative Affect	b = −0.43, t(87) = −0.93, p = 0.35	b = −0.39, t(86) = −0.87, p = 0.38	b  = −1.23, t(86) = −2.79, p = 0.006**
IOS	b = 0.33, t(86) = 1.43, p = 0.15	b  = 0.59, t(85) = 2.55, p = 0.01**	b = 0.24, t(85) = 1.08, p = 0.28
Connection (Group)	b = 0.33, t(86) = 1.38, p = 0.16	b = 0.15, t(85) = 0.6, p = 0.54	b = 0.12, t(85) = 0.52, p = 0.6
Perceived Pain	b = −0.25, t(86) = −1.25, p = 0.21	b = −0.14, t(86) = −0.73, p = 0.46	b = 0.08, t(86) = −0.42, p = 0.67
Cortisol	b = −0.002, t(39) = −0.09, p = 0.93	NA	NA

Note: This table shows corresponding coefficients from multilevel models of each outcome measure. * p < 0.05, ** p < 0.01, *** p < 0.001

This table also shows coefficients of the interaction terms between Session and Condition across each outcome measure. As shown, none of the interactions were significant, suggesting that differences in changes between synchronous audio and muted conditions did not differ across sessions across all outcome measures.

Negative Affect (NA)

Similarly to trends in positive affect, participants reported significantly lower levels of negative affect after the singing activity in comparison to their pre-singing state (d = 0.39; Figure 3A). This indicates a modest but reliable reduction in negative affect following group singing. When examining whether these changes varied by session or singing condition, no significant interactions emerged (Figure 3B; Table 2). There were also no differences observed between singing conditions. Although there was no difference in changes in negative affect across session 2–4, when collapsing across singing conditions there was small but significantly larger reduction (d = 0.28) in negative affect scores in session 5, compared to 2. Taken together, these findings indicate that group singing produces reductions in negative affect and that these reductions may increase over time, but do not vary between singing conditions.

Figure 3.

Violin plots showing changes in negative affect. (A) Negative affect scores measured before (pre) singing sessions were significantly different from those measured after (post) singing sessions, p < 0.001. (B) Change in negative affect (post-pre) did not differ significantly between audio conditions, when collapsed over sessions. Points in the violin plots represent means, and error bars represent bootstrapped 95% confidence intervals. Both are generated by the ggplot function stat_summary, which by default uses 1000 repetitions to determine confidence intervals.

Inclusion of Other in the Self (IOS)

Results revealed that participants reported significantly higher levels of IOS after the singing activity in comparison to their pre-singing state (d = 0.41; Figure 4A). This indicates a modest but reliable increase in IOS scores immediately following group singing. When examining whether these changes varied by session or singing condition, no significant interactions emerged (Figure 4B; Table 2). There were also no differences observed between singing conditions. Although there was no difference in changes in IOS scores in session 3 and 5, when collapsing across singing conditions there was small but significantly larger reduction (d = 0.26) in negative affect scores in session 4, compared to 2. This suggests a small, albeit reliable, benefit of prolonged group singing. Taken together, these findings indicate that group singing produces increases in IOS scores and that these benefits may accrue over time, but do not vary between singing conditions.

Figure 4.

Violin plots showing changes in responses to Inclusion of Other in the Self (IOS) scale. (A) IOS scores measured before (pre) singing sessions were significantly different from those measured after (post) singing sessions, p < 0.001. (B) Change in IOS scores (post-pre) did not differ significantly between audio conditions, when collapsed over sessions. Points in the violin plots represent means, and error bars represent bootstrapped 95% confidence intervals. Both are generated by the ggplot function stat_summary, which by default uses 1000 repetitions to determine confidence intervals.

Connection to the Group

Like IOS scores, results revealed that participants also reported significantly higher levels of Connection to the Group in measurements following singing compared to prior (d = 0.37; Figure 5A). This indicates a modest but reliable increase in Connection to the Group scores immediately following group singing. However, unlike IOS scores, Connection to the Group scores were shown to be significantly higher in the synchronous audio, compared to the muted audio group (d = 0.28; Figure 5B). This suggests a small but reliably higher increase in Connection to the Group following synchronous group singing, compared to the muted audio singing. There were no significant changes across sessions and no interaction between singing condition and sessions (see Table 2). Overall, these results suggest that regardless of audio condition group singing increases feelings of connection to the group, but this increase is significantly larger when individuals share auditory information.

Figure 5.

Violin plots showing changes in connection to the group. (A) Reported connection to the group measured before (pre) singing sessions was significantly different from those measured after (post) singing sessions, p < 0.001. (B) Change in connection to the group (post-pre) was significantly different between audio conditions, when collapsed over sessions, p = 0.02. Points in the violin plots represent means, and error bars represent bootstrapped 95% confidence intervals. Both are generated by the ggplot function stat_summary, which by default uses 1000 repetitions to determine confidence intervals.

Reported Pain

Results revealed that participants reported significantly lower levels of reported pain after the singing activity in comparison to their pre-singing state (d = 0.14; Figure 6A). This indicates a small but reliable decrease in reported pain immediately following group singing. When examining whether these changes varied by session or singing condition, no significant interactions emerged (Figure 6B; Table 2), and no differences observed across sessions. Taken together, these findings indicate that group singing, regardless of the auditory condition or the number of sessions attended, reduces levels of reported pain.

Figure 6.

Violin plots showing changes in reported pain. (A) Reported pain measured before (pre) singing sessions were significantly different from those measured after (post) singing sessions, p < 0.04. (B) Change in reported pain (post-pre) did not differ significantly between audio conditions, when collapsed over sessions. Points in the violin plots represent means, and error bars represent bootstrapped 95% confidence intervals. Both are generated by the ggplot function stat_summary, which by default uses 1000 repetitions to determine confidence intervals.

Cortisol

Results revealed that participants reported significantly lower levels of cortisol in the measurement following group singing, compared to prior (d = 0.21; Figure 7A). This indicates a small but reliable decrease in levels of cortisol immediately following group singing. When examining whether these changes varied by session or singing condition, no significant interactions emerged (Figure 7B; Table 2). There were also no differences observed between singing conditions or across sessions. Taken together, these findings indicate that group singing, regardless of the auditory condition or number of sessions, reduces biomarkers of stress through reductions in levels of cortisol.

Figure 7.

Violin plots showing changes in salivary cortisol. (A) Salivary cortisol concentrations measured before (pre) singing sessions were significantly different from those measured after (post) singing sessions, p < 0.001. (B) Change in salivary cortisol (post-pre) did not differ significantly between audio conditions, when collapsed over sessions. Points in the violin plots represent means, and error bars represent bootstrapped 95% confidence intervals. Both are generated by the ggplot function stat_summary, which by default uses 1000 repetitions to determine confidence intervals.

Qualitative Results

The five major themes that emerged in our content analysis of codes and categories from focus groups (FG1 and FG2) as well as the individual interviews (II) were as follows: 1) the impact of technology modality, 2) the sense of social connection through musical engagement, 3) the power of music, 4) the health benefits of embodied singing, i.e., singing with conscious awareness of bodily engagement, and 5) the pedagogical facilitation aspects of the class. These themes, and their sub-themes, are shown in Table 3.

Table 3.

Qualitative themes.

Theme	Subthemes
Theme 1: Impact of Technology Modality	Muted audio platform was a bit lonely, less connection
	“People in boxes” on the videoconferencing platforms
	Muted audio enabled opportunity for increase in confidence
	Synchronous audio system installation transition brought anxiety and technology challenges
	Synchronous audio enhanced their experience, more like in-person
	Technology issues seen as positive/learning experience and brought further connection/cohesion
	Blending of voices enhanced the group's musical experience
Theme 2: Social Connection through Musical Engagement	Synchronous audio increased sense of connection to one another “Being known”, with deeper collective connections through sharing of stories
	“Encouraged by others' creative expression”.
Theme 3: The Power of Music	Singing impacts mood
	Music elicits expression and release of emotions
	Music as a memory maker and memory keeper
	Music as language
Theme 4: The Health Benefits of Embodied Singing	Embodied singing/visceral experience
	Breathing as a practice
Theme 5: Pedagogical approach, tech support, and course length.	Role of the facilitator
	Meaningful song choices
	Technology support was an essential ongoing component
	Desire for longer course duration

Theme 1: Impact of Technology Modality

Within the range of topics regarding the muted audio modality, an emergent subtheme was the sense of subdued connection among participants when they could not hear each other singing. One participant described singing in the muted audio format as being “more like a private lesson as opposed to a kind of communal experience” (FG1, participant 2) while others spoke of how they were “…aware of all the people but didn’t really…feel very connected at all during the first [sessions]” (II, participant 17) and “…not being able to have one collective voice, it does feel like something's missing” (FG1, participant 1). Several participants spoke of “people in boxes” or “little squares,” referring to how everyone looks on the videoconferencing software. As expressed by participant 13, “I’m sitting in a room with these cardboard cutouts. I can see mouths moving and people swaying but, because I can’t hear them, I don’t particularly feel very connected to the group” (FG1).

In an interesting counterpoint, a related perception voiced by many participants was that they found a certain liberation in not being audible to others at the outset of the study, appreciating the span of muted preliminary sessions in which to build courage and confidence in singing skills before the transition to a synchronous audio platform. “I [am] more comfortable for now singing without other people hearing me…but it will be good to be able to hear one another after [I am] more familiar with the songs” (FG1, participant 11). Supporting that insight, participant 5 described the muted singing sessions as an opportunity to learn more about their own voice, its abilities and limitations, as they navigated their lack of voice control due to a neurological disorder (FG1).

In contrast, when discussing the synchronous audio sessions, all but one participant offered extremely positive comments regarding the enriching nature of their overall experience, despite initial frustrations and anxieties during the installation process for the low-latency access units and/or app, and the preliminary engagement with the unfamiliar platform. Participant 12 noted that after the technology issues were resolved, she loved “every part of it…the idea of it, the community of it, how I felt after it, was magnificent” (II). An intriguing artefact of the difficulties encountered when adapting to the new technology was that the shared challenges were identified as a factor in strengthening the group's connection, which “…continued to grow, definitely because of our ability to sing together and hear each other but also I think partially because of the challenges we had with it. That feeling of sympathy for the others who were struggling to get online and participate. I think that also increased our bond together as a group” (II, participant 1).

Participants’ responses indicated emphatic enthusiasm regarding the ability to hear harmonies on the synchronous platform: “I could hear some people who were doing harmony with the songs. That was wonderful. That was a big breakthrough. Because before it was you were just doing it basically by yourself, with one or two people that you could hear, so I really found that was marvellous…. It is so fun to be inside that sound world” (II, participant 6).

Theme 2: Social Connection Through Musical Engagement

The sense of being known through the sharing of both stories and songs presented as a powerfully compelling theme across all qualitative content. In referencing the muted sessions, a participant observed: “And I think [that] the singing with other people, even though we can’t hear each other, we’re doing something all together at the same time so there's this sense of shared experience even though I don’t feel very connected…a kind of reassurance that we can all be supporting each other even though we don’t really know each other beyond the name that's on the square yet” (FG1, participant 13). The experience of hearing one another led to a deeper sense of connection compared to the muted platform: “[I] definitely feel a lot more connected through the [synchronous audio] technology,…hearing the sound of the collective sound is a lot more rewarding than just knowing I’m in isolation here and other people out in the ethernet are singing the same thing I’m singing even though I can’t hear them. It's sort of like mono compared to stereo times ten” (FG2, participant 1).

Connection between participants also grew as they shared emotional memories awakened through singing. Reflecting upon a tender moment of responding emotionally to a song that connected them to their past, participant 17 expressed: “Some people might see my tears anyway, so I might as well explain where they’re coming from. That actually kind of took a little bit of a step of courage I guess, and I also know that I’m known now” (FG1). Articulated another way, “…when you tell your story…it connects me to you” (FG1, participant 8). Several participants mentioned that connections grew after the first focus group as it created an opportunity to get to know one another more directly and to share stories and experiences. For example, participant 13 noted, “I think for me it really changed when we had the first focus group, got to hear people's voices, got to hear them say their own name, and not just read it on the screen” (II).

In sessions conducted on the synchronous audio platform, participants felt encouraged by each other’s musical and creative expression. Many noted both the beauty and sense of cohesion of participating in songs sung in a round [canon], as encapsulated here: “When we did the round with ‘Viva La Musica’, that was one of those instances of ‘my group has to do [our] bit. I rely on the other two groups to do their bit.’ And we have to work together, we have to listen to one another, we’re doing different things together” (II, participant 17).

Theme 3: The Power of Music

Beyond the theme of music-based connection, several sub-themes regarding the power of music were identified within the participants’ commentary. Many observed the impact of singing upon mood, and its innate nature as a venue for natural expression and release of emotions. As well, participants reflected on music's capacity as a form of language, and its unique function as a “memory maker” and “memory keeper.”

The act of singing, the song lyrics, and being part of the choir were noted by many as a source of joy: “I mostly just feel joy when I’m singing. Singing makes me feel joy” (FG1, participant 16). In describing how his mood was affected beyond the sessions themselves, participant 17 indicated that “…sometimes the songs will be in my heart, um, and they’ll be going through my head for quite a while and that definitely, when that happens, my mood is higher too, I’m in a better mood for sure” (FG2). Singing was also seen as a way for some participants to put aside the worries and difficulties in their lives: “It reminds me to count my blessings and not always dwell on the sad things that are going on in the world now. And with ‘Thank You for the Music’, the singing just really cheers me up” (FG1, participant 11). Music can evoke a range of emotions: “Those first few times we would start singing that song I would literally tear up. So that's the one that evokes emotion” (FG1, participant 16).

Across all qualitative discussion forums, music's intriguing capacity to function within the brain as a guardian of personal memories was a consistent topic of reflection. Connecting with songs from the 1500s or the 1950s was described as “time-travel” by participant 7: “So, she's brought some pieces in that I sang in my high school choir or took me back to when I was a child and, I know it's that time travel aspect again, but it's recreating something from a different part of your life and I’ve found that to be very meaningful over my time in this class” (FG2). The contemplation of music as a special and distinct form of language was expressed by participant 13: “it's some kind of deep seated inner creativity, inner expression that needs to come out that words don’t capture so music can be that expression… But music, I think in that way, can express things that are beyond words, that words aren’t really necessary, but that the emotion or the sentiment, the feeling, the desire can be expressed um, through music” (II).

Theme 4: The Health Benefits of Embodied Singing

Engaging in song, in combination with the practice of intentional diaphragmatic breathing and postural alignment provided an opportunity on both platforms for participants to experience a profound level of connection to their own inner world. “I really like how [our director] has us moving our bodies, like, I’ve never done that in any other choral group before, where there's a lot of focus on breath and moving our bodies and I just think that that's bringing in the senses, it's just allowing our emotions to move through us and the sensation to just move through and to just be able to flow with whatever feelings that the music is bringing up for us” (FG1, participant 7). Participant 2 noted “after the last session on Wednesday, I was feeling really good, like, physically good,… And it's because of the music and the singing and being able to belt it out and, I guess, getting all of that oxygen in” (FG1).

Theme 5: Pedagogical Approach, Tech Support, and Course Length

Participants noted that the skillset implemented in the music-based facilitation guided their own narrative and creative connections to the music, fostering and sustaining the sense of communal experience as participants collectively shifted with ease within levels and states of shared expression throughout the study. Participants appreciated the varied structure and activities of the class, as noted by participant 7: “I really love the format that [the facilitator] has used all along. I love that she brings in the movement like being in the garden, you know, the chimes, just all of the things to just engage your mind and your body… I love the movement part of it, the breathing part of it, the time to reflect at the beginning and then at the end… I just think that those elements actually for me have been…meaningful and powerful” (II). Repertoire choices were also noted by many as well chosen to impart importantly meaningful content. “The songs were, I think she chose the songs beautifully, because they were very emotionally charged, and it brought us together” (II, participant 12).

The group was very thankful for the technology support provided by the student research assistant team, and indicated how essential they were to the platform transition process: “I was one of the people that experienced more trouble than others, so just to have that, I mean, the support was great…they figured out work arounds and they had eventually ways for people to participate” (II, participant 13).

Overall, participants wished the course had a longer duration, so that they could have had more sessions using the synchronous audio platform once technological issues had been solved. This was clearly expressed by Participant 5, who said, “We’ve gone through all this learning and all this stuff, we should be carrying on and enjoying the singing!” (FG2).

Ethnographer Observations

The team's ethnographer, whose research as an ethnomusicologist explores affect and intersubjective experience, conducted focus groups I and 2 (FG 1 and 2) and the Individual Interviews (II). She reported strong experiences of connection during FG 1 and 2 and II, wherein participants displayed affectual markers of camaraderie and connection such as smiling, tearing up, supporting one another's statements, and sharing personal histories and feelings. During a feedback session with the research team, the ethnographer suggested the focus groups and Individual Interviews were a significant aspect of the intervention, in addition to their value as occasions for gathering qualitative data. FG 1 and 2 provided a platform for exchange and reflection among the participants that engendered feelings of connection, with several participants noting that they appreciated being introduced to one another and speaking more directly. Participants cited multiple benefits of participation in the study, expressing a strong desire for an extension of the class sessions, information regarding the quantitative data analysis (the cortisol level in saliva samples, for example), or having the opportunity to participate in future singing collectives. As participant 13 expressed, “…doing something creative and fun like singing, and doing it in a group with someone as compassionate as [the facilitator] and as encouraging as she is…is a real boost and a real encouragement to keep going, no matter what is happening in your life so, it's a nice little pause. And then, it gives a boost of energy and strength to carry on” (FG2). They later noted “Part of the joy of being able to hear other people is being inspired, whether it's a harmony, whether, you know, there's one song where there's a line that some people sing and, whether or not you sing it, it just kind of sparks a different kind of enjoyment of making music together.” (FG2, participant 13). In FG 2, Participant 16 offered a concise summation of the journey: “The whole experience has been an absolute pleasure.” The ethnographer observed that the positive experiential outcomes, as evidenced in the qualitative and quantitative data, supported the study's multi-disciplinary methodology and knowledge continuum as applied to the facilitation design and implementation.

Comparison Between Quantitative and Qualitative Results

Both sets of results show that participating in online singing in both audio formats leads to improved mood after each session compared to before the session. The quantitative results show a trend towards a greater increase in positive mood during the synchronous audio sessions compared to the muted audio sessions, and this is well-supported by the qualitative results, which indicates that sharing in the blending of voices enhances the singing experience and leads to joy.

The quantitative results show that feelings of social connection are increased by singing in an online group, and both sets of results support the idea that the synchronous audio condition leads to a greater sense of connection, with the qualitative results adding the context that the participants felt supported and encouraged by being able to hear the others singing. In addition, the qualitative results indicate that sharing personal stories in between songs, on either platform, contributed to the development of social connection.

Overall, both the quantitative and qualitative results of this mixed-methods study point to some areas of elevated benefits within the synchronous audio platform experience, while also acknowledging that different benefits within the non-synchronous modality were also cited by participants. The quantitative results are more understated, while the qualitative results provide richness and context, allowing us a deeper understanding of how the different elements of the singing sessions (e.g., breathing exercises, sharing stories, songs that evoke memories) each wove together to contribute to the overall effects on mood and connectedness.

Discussion

Our study demonstrated that facilitated virtual singing groups can benefit participants’ sense of connection as well as both physiological and psychological wellbeing. Both modalities of virtual sessions led to decreases in salivary cortisol, improved mood, decreased pain, and an increased sense of connection amongst the singing group participants and to the world at large. Being able to hear others singing and be heard by them in turn in the synchronous audio condition led to marginally greater increases in positive mood and significantly greater increases in the sense of connection compared to participation in the muted singing condition. Beyond these two observations, there were no other differences between the two conditions in quantitative outcomes.

To summarize our qualitative data, direct participant commentary spoke to a range of nuanced experiences, especially as to the complexities of adapting to a synchronous platform that offered challenges but also enhanced musical possibilities. Participants noted that singing on the muted audio platform was enjoyable and encouraged individual confidence regarding personal vocal capacities. However, participants observed that the shared element of social connection was slow to evolve in that setting. While initial implementation of the low-latency audio platform brought technological frustrations, the shared challenges may have galvanized deepened feelings of connection to the group. The ability to hear the collective singing voices greatly expanded the musical experience. Overall, shared engagement through songs and stories was seen as a powerful way to express emotions and awaken memories through song and story. Participants noted health benefits from the class, associated with deep breathing and the embodied experience of singing.

Our results contribute to the expanding domain of research and practice, providing advocacy for group singing as a beneficial resource in multiple areas of health and wellbeing activities. Studies investigating the physiological responses to singing have quite consistently shown decreases in salivary and serum cortisol levels, indicating lower stress levels (Beck et al., 2000; Fancourt et al., 2016; Kreutz et al., 2004; Mallik et al., 2025), and our results are in line with those findings.

While our results indicate that both cortisol levels and reported pain benefit equally from muted audio online group singing and low-latency synchronous group singing, the effects on mood responses trend toward an increased benefit of synchronous singing, suggesting that the social and collaborative artistic co-creation elements of group singing have an added effect on mental and emotional wellbeing. Assessing session-by-session changes in positive affect (Figure 2), effects of muted audio group singing are shown to be consistent across sessions. In contrast, changes in positive affect during the synchronous audio group singing increased over the sessions, with a maximum in the fourth session. Drawing from multiple participant narratives, we speculate that the increase in positive affect was due to the enhanced social connection catalyzed as an outcome of hearing one another synchronously, and the resultant sharing of artistic expression and musical joy in the virtual collective singing space.

In earlier synchronous audio sessions, this effect was likely tempered by the technological challenges of becoming comfortable with a new platform. Paradoxically, the change in positive affect is diminished in session 5 of the synchronous audio condition, which was the final session; the mean change in positive affect decreased from 6.5 in session 4 to 2.5 in session 5. Based on participant's comments in the qualitative interviews regarding their mood in the final session, this smaller change in positive affect was due to participants’ feelings of sadness and loss in confronting the conclusion of their time together and is consistent with the qualitative subtheme that participants felt the course end came too soon. This decrease in the effect of positive affect is therefore actually indicative of a positive outcome: joy in the experience of shared song across distinct virtual spaces. Synchronous audio group singing offered study participants a deeply meaningful shared artistic experience and there was a sense of loss that the course was ending. As noted by participant 1, “It's been great. It's sad it's coming to an end so soon… It really has been uplifting to join in with you others… It's been a blast!” (FG2). We hypothesize that if the study had continued for more sessions using the synchronous audio platform, positive affect changes would have plateaued at a higher level, and we would have seen a significant difference in positive affect between the two platforms. This remains to be confirmed with future studies, but in the words of participant 18: “This class is a real blessing… I feel every time after the session, I feel a burst of energy. It's a miracle.”

The correspondence between mood and social connection is supported by comments from participants. When reflecting on his mood during the synchronous sessions, participant 1 expressed how “It definitely was a great psychological uplift having that participation with other people.” Participant 18 described the experience of synchronous singing as follows: “Each of us has a struggle to deal with, like a challenge, but each of us can be together and transform the negative energy, you know, into collective positive energy.” This idea comports with the hypothesis that group singing has a strong effect on social cohesion, and leads to decreases in loneliness (Kim & Kim, 2024). Participants engaged in group singing report strong feelings of connection with their peers, variously reported as social cohesion, belonging, or relatedness. (Lonsdale & Day, 2021; Pearce et al., 2015; Weinstein et al., 2016). Social isolation, increasingly common among the elderly and people with health issues, is a root cause of low mood (World Health Organization, 2025). The ability of group singing to increase social connection and thereby improve mood is a key element of the therapeutic lift that comes from making music together. Online group singing in a muted format continues to offer social connection benefits, and for some, it is a preferable meeting environment, offering a greater sense of personal privacy while participating. While individuals can mute themselves in the synchronous environment, identifying that form of non-participation in that shared environment may generate other problematic personal elements. For those who are able to participate in unmuted synchronous audio, our results suggest that the format engenders an elevated perception of group cohesion.

This study was designed as a pilot, to test the feasibility of using low-latency audio with a mixed population in a therapeutic setting, and to begin to understand the difficulties and advantages of this technology compared with the more commonly used high-latency alternatives. Our results are limited by our small sample size, and by the restricted number of classes conducted using low-latency audio, which gave the participants limited time to get comfortable with the technology before the end of the study. We chose to study a mixed group of participants; this naturalistic approach therefore precluded a study design which would identify therapeutic outcomes associated with a particular population. The non-specific participant focus was a deliberate aspect of the study's areas of pedagogical interrogation, exploring facilitation methods that invite patients to participate in a broad community rather than being defined by their medical conditions. This inclusive approach is also more representative of the broader community of older adults.

Our study design did not include a separate comparison group. Because the participants completed seven classes together using a muted audio platform prior to the introduction of the low-latency platform, the effects of social connection in this second modality are confounded by the fact that the participants had already developed a familiarity with each other. Instead, our design allowed us to chronicle the transition between the two modalities, and the qualitative results are strongly in support of a benefit of the synchronous audio technology in promoting social connection among participants. An emergent logistic difficulty with our study is the considerable challenge of collecting quantitative data remotely. The participants were reminded to complete online questionnaires and collect salivary samples before and after classes and were generally, but not always, consistent in performing these tasks in a timely manner. We were able to track the times at which the questionnaires were completed and found that the median time to completion of the post-class questionnaire was 15 min, with 20% of the questionnaires being completed more than 1 h after the end of the class. The timing of collection of salivary samples was not recorded. The delay in collection of questionnaire data and salivary samples likely led to increased variability in our quantitative results, which nevertheless did capture significant differences in pre- and post-session measures.

Video conferencing technology is in a state of rapid advancement. When we began planning the study in 2022, most low-latency platforms required dedicated hardware devices to ensure reliable performance. During the course of the study, software-only solutions were introduced for MacOS users. We were able to incorporate that option for those with access to the necessary operating systems. However, the implementation of the synchronous audio platform involved some technological challenges. Research assistants visited each participant's home to install software and hardware, which required some problem solving. Each participant's home setup was unique, which meant that every installation was slightly different; internet infrastructure, operating systems, and hardware all varied widely. All users required wired access to the internet via Ethernet. Challenges included the location of the internet router with respect to the participant's computer (requiring extra-long Ethernet cables in some cases), inconsistent internet speeds, outdated operating systems, or microphones that were too quiet. Because of these issues, and because we wanted to ensure that everyone could join synchronously without frustration, we decided to delay the start of the synchronous sessions. Session 7 was delivered via the muted audio platform so that we could finish troubleshooting with the remaining participants and stabilize everyone's connections before beginning synchronous sessions.

There are ongoing developments in the field of synchronous audio technology, with multiple networking and communications platforms exploring low-latency audio integration. Software-based low-latency audio options are available on a range of operating systems, opening an increasingly wide range of access for people interested in experiencing this modality. As this technology improves, we expect to see low-latency audio options for online choirs become more readily available and accessible. Despite these advances, at the time of publication there remain some inherent access barriers to broader use of low-latency audio platforms in this context including the cost of equipment and subscriptions, requirement for users to be within a somewhat restricted geographical area (∼250–500 miles from each other and the central server), knowledge of software and hardware setup, and the need for high-speed wired internet connections.

Our study aimed to understand the advantages and disadvantages of online group singing, and how these are impacted by synchronous audio technology. On the muted audio platform, holistic elements of musical care and a sense of connection to the director arose due to the sense of common artistic and social goals supported collaboratively. The theme of acknowledged limitations of muted audio for virtual group singing was central to the first modality of engagement while also enabling the participants to prepare for the shift to the synchronous audio platform. Relatedly, an unanticipated and important benefit of using the muted platform was the level of comfort for some participants who felt self-conscious about their singing voice being heard by others. Participant 18 identified a sense of freedom in the experience of vocally expressing themselves without a worry of disrupting the collective sound: “I was feeling a little bit nervous about my voice quality…and so I did appreciate…the nature of the sessions that allowed me to sing without being exposed.” Another element of connecting with one another on the muted platform came through the opportunity to engage audio for dialogue about the repertoire. In the words of participant 12, “there was a difference, there was a definite noticing when we could share stories with each other, like when we actually started to communicate with each other or somebody would have an opinion about something, of course that connected us much more and it became much more interesting then.”

Even the expression of hand symbols for vocal vowels while singing on the muted platform gave a sense of synchronous activity to the participants. The facilitator incorporated this pedagogical method to enhance participant understanding of palate shape when pronouncing the vowels, but it also provided a collective expression.

As described by participant 16 when discussing the muted sessions, “The biggest connection was when we were doing the hand symbols together. …singing on our own, usually I’m just looking at [the choral director]. I don’t even have the whole panel in front of me so I’m not watching [others] during the part where we’re singing, looking at the music and trying to learn the melody and the tempo and everything. But…I’ve felt more of a connection…during the time that we’re doing that Solfege and afterwards” (FG1). There is a depth woven into such an embodied exercise, even when there is no audio capacity to hear one another, revealing alternative ways to feel connected.

This participant's description of synchronized participant hand gestures as a collaboration indicates how that action piece not only enhanced the collective's pronunciation of vocal vowels but also granted a moment of shared creative expression. Approaches such as this, intended to foster both care and trust amongst the entire participant and facilitation team, can provide substantive instruction regarding tools of musical craft. As the virtual singing space and related music wellness practice continues to evolve, a new realm of creative virtual music pedagogy and collective participation is poised to emerge, with immense potential expansion of beneficial applications.

As they each transitioned from one modality to another, some participants spoke to the shared opportunity of caring for each other and encouraging one another – and how that experience became centrally important to their increased sense of connection. In describing the shift to synchronous audio technology during her individual interview, participant 13 shared how the frustration and challenges created lessons of patience and kindness toward self and others: “I felt a deeper sense of how we’re all in this together…we can do this together” (II). This supports both the qualitative and quantitative finding of a bond that developed within the singing collective, in which the logistical obstacles conversely enhanced group cohesion and empathy.

Participant representation in the study's singing group of a multi-strand population required an interpersonal ethos of care extending from the musical facilitator's model to encompass the participants’ collective interactions. Specialized content design took into consideration the specific needs and hopes of all participant cohorts: older individuals living with diverse neurological conditions, care partners, and healthy agers. For the patient cohort, the sense of not being defined by one's condition and yet knowing that it was communally acknowledged fostered a sense of trust and respect across the collective. During one of the mid-study focus groups, participant 1 shared about his personal health pathway in the context of the singing collective: “One of the benefits for me in participating is that I get a lot of breathing exercises which is very, very valuable to me with my condition…using my lungs, my voice, my throat muscles, my vocal chords to sing is so much more enjoyable. […] You also just emotionally feel so much better because we’re doing it as a collective even if we can’t hear each other yet. There's still that happiness, that warmth inside. It's not quite a tingle but you know what I mean. So, you feel part of a good community, a shared community.”

Such shared experiences throughout the study made it possible for the facilitator and participants to engage with six concepts of lifespan learning: those of connection, creativity, curiosity, cohesion, collaboration, and care – elements that are grounded in relational developmental systems theory (Lerner et al., 2015; see Friesen, n.d., for details of how these concepts were integrated into the pedagogy of this singing collective). Such an approach to virtual music making, especially when intended as a health-arts lifestyle intervention, could inform the paradigm of practice for music facilitators by illuminating the capacity of everyone involved to collaborate and learn together.

While the initial impetus for the exploration of synchronous virtual audio in therapeutic facilitations was mobilized by the potential for expansion of group-singing capacities for participants, it was designed with an understanding that it would naturally lead to concurrent explorations of an expanded range of facilitation methods. Perceiving both individual and collective needs in virtual spaces, observing and listening clearly through digital devices, requires a dedicated intentionality driven by the moving conviction that music has the power to connect us all deeply. We have not detailed those aspects in this paper but suggest that they merit further exploration.

Conclusion

This study of synchronous, low-latency audio was an early pilot exploration of potential applications for therapeutically oriented and facilitated group-singing activities. While our quantitative results indicate subtle benefits of synchronous audio group singing compared to muted audio, the qualitative findings paint a stronger picture. The modality transition from muted to synchronous audio technology shifted its participants into a shared space wherein a new form of real-time collective singing became a reality. As described by participant 17, the group's co-created harmonies sparked a joyous sense of expanding communion: “Any time where we had some kind of harmony going on helped to…have that sense of…independence and unity simultaneously, that's the beauty of it. That's what harmony does, so that makes me feel more connected.” For people who enjoy group-singing activities but are chronically isolated through medical, mobility, remote location, or other circumstances, synchronous audio technology provides an access portal to the profound power of singing together. While our pilot study was small in scope, we hope that the meaningful experiential outcomes narrated by its participants will support impetus for further explorations within the emerging virtual realm of therapeutic group-singing activities.