Acoustic monitoring in care homes: how it works, what it hears, what it misses

How does acoustic monitoring work?

Two main designs. Both run at night, and a person still decides whether to go in.

Threshold systems. The older design. CLB says the approach "has been around for more than two decades", and that when "any sound profile exceeds its individually set threshold, an alert is sent to a central station or forwarded to a mobile device" (CLB, accessed 2026-09-05). Each resident gets their own level. A loud snorer gets a higher threshold than a quiet sleeper.

Sound-class systems. These try to name the sound. Earzz says customers choose from a catalogue "currently hosting more than 20 sounds", listing Thud, Scream and Cough among them, and that notifications "are usually delivered within 5 seconds of the sound event once recognised although this may vary depending on network" (Earzz, accessed 2026-09-05). Ally says staff can "review sensor data, listen to audio samples, and assess whether residents need their assistance" (Ally, accessed 2026-09-05).

What can it hear?

A call for help. Crying or shouting. Someone getting out of bed. Anything loud enough to cross the line set for that room.

The case for it rests on sleep, not falls. Night checks wake people. In four nursing homes, noise and light changes from the ward environment and from nursing care were linked to 50% of waking episodes of four minutes or more among 118 incontinent residents, and "the major sources of all noise were traced to nursing staff" (Schnelle et al., J Am Geriatr Soc, 1993).

The SIESTA authors write that "nearly 40% of care home residents who are living with dementia also have symptoms of disturbed sleep" (Webster et al., 2020). They cite this as background from earlier work, and name no country. In their interviews, staff said daytime sleepiness "led to falls, missing medication, drinks, and meals".

One caution. A later randomized trial in eight nursing homes with 267 residents cut the two-minute intervals per night with peak noise above 50 decibels from 83 to 58. The authors still concluded that this "did not lead to significant improvements in the day sleep and most night sleep measures" (Schnelle et al., 1999). A quieter ward did not, by itself, mean better sleep.

What does it not hear?

Silent events. A slow slide from a chair to the floor. A resident who has not moved in hours. Someone leaving the room quietly at 2 a.m. None make much noise, so none reliably trigger an alert.

Accuracy, in a vendor's words. Earzz states its product "may not always be accurate or reliable in recognising sounds" and "should not be relied upon in high-risk or emergency situations" (Earzz, accessed 2026-09-05).

Alarm fatigue. In three English dementia homes, monitoring tools "generated frequent alarms that placed a burden upon staff" (Hall et al., Int J Nursing Studies, 2017).

What does the evidence show, and how strong is it?

The headline numbers come from a Care England report produced with the vendor Adaptive Care. It states that across six Quantum Care homes "nearly 94,000 physical night checks and over 8,000 staff hours were required over a three-month period under traditional models", and that after acoustic monitoring "routine room entries reduced by 90%". It reports a "74.5% falls reduction", of which 64.5% were true falls and 9% incidents thought to have been a fall that led to a 999 call out (999 is the UK emergency number; 911 in Canada). No home is named for that figure (Care England with Adaptive Care, 27 March 2026). These are claims, not peer-reviewed findings, and the page publishes no method. We did not retrieve the full PDF.

WCS Care says it began its trial in 2016 and that "within a year we evidenced a 34% reduction in night-time falls", plus a 55% reduction across days and nights (WCS Care, accessed 2026-09-05).

Vendor case studies are small and short. Ally's six-home pilot with NHS Dorset Integrated Care Board reports a 49.2% cut in total falls, but notes "one site showed an increase in total falls" (Ally Cares, accessed 2026-09-05). Earzz's Terrington Lodge pilot compares 13 falls at baseline with 2 over five weeks, with no control group (Earzz, accessed 2026-09-05).

For Canadian scale: 72,392 fall-related hospitalizations among people aged 65 or older in 2019/2020. Over 2015/2016 to 2019/2020, 52% of hospitalized falls happened in a household residence and 17% in a residential care facility, with no time-of-day breakdown (PHAC, 2024).

Is audio recorded, and who can hear it?

It depends on the vendor.

• Replay, and possibly video. CLB says that with an alert "if available, a video connection can be made", and that "alert-triggering events can be replayed if required" (CLB, accessed 2026-09-05). Ask whether video is enabled in your home.
• Audio samples and transcriptions. Ally's product privacy policy lists "audio signal data" and "transcriptions of audio data" (Ally Labs, accessed 2026-09-05).
• Sounds, not speech. Earzz says it recognises "just sounds, NOT SPEECH" (Earzz, accessed 2026-09-05), and describes "no recordings, no microphones accessible by humans, and fully anonymised triggers" (Care England, Earzz post, 31 July 2025).

A bedroom microphone also captures staff, visitors and roommates. Several Canadian laws may apply.

• Federal. Where PIPEDA applies, "the knowledge and consent of the individual are required" (Office of the Privacy Commissioner, 2025).
• Ontario. The Fixing Long-Term Care Act, 2021 includes a Residents' Bill of Rights. PHIPA (2004) sets rules for personal health information.
• British Columbia. The Personal Information Protection Act (2003).
• Alberta. The Health Information Act.

Ask the home's privacy officer which one applies.

How does it compare with other room sensors?

Sources for the acoustic row are above. The other rows describe these sensor types in general, not any vendor.

What should you ask before buying?

• How is resident consent handled, including substitute decision-makers?
• What is stored, for how long, and in which country?
• Is there Canadian support and hosting? We checked the public websites of CLB, Ally and Earzz on 2026-09-05, and none states Canadian availability, hosting or support.