Speech Intelligibility and STI: Making Sure Every Word Is Heard in Meeting Rooms

Speech Transmission Index (STI) is a 0–1 score that predicts how much of what someone says will actually be understood in a given room. In a meeting room the target is high STI — above roughly 0.60, ideally above 0.75 — achieved by keeping reverberation time around 0.5–0.8 s, controlling background noise, and maintaining a strong signal-to-noise ratio. When STI drops below about 0.45, listeners start missing words, asking people to repeat themselves, and struggling on video calls.

If you have ever sat in a glass-walled meeting box and found yourself leaning forward, asking “sorry, can you say that again?”, you were experiencing low speech intelligibility firsthand. It is one of the most common — and most fixable — problems we encounter in our site surveys of corporate meeting and conference rooms. This article explains what STI actually measures, what damages it, how it is tested, and what we do to bring a failing room back into a usable range.

What is Speech Transmission Index (STI)?

STI is a standardized, objective metric — defined in IEC 60268-16 — that predicts how intelligible speech will be after it has travelled through a room (or a sound system, or both) to a listener. It does this by measuring how much a test signal’s modulation pattern — the rise and fall of speech energy over time, which is what carries phonetic information — gets smeared or masked between source and receiver. The more the modulation gets flattened by reverberation, or buried by background noise, the lower the STI score and the harder speech is to understand.

The scale runs from 0 (completely unintelligible) to 1 (perfect intelligibility). IEC 60268-16 defines the qualitative bands most acousticians use in conversation:

For a typical corporate meeting room, we treat 0.60 as the minimum acceptable target and 0.75+ as the goal for rooms that host client presentations or heavy video-conferencing.

Why does a meeting room need HIGH speech intelligibility (unlike open-plan)?

This is the point that trips people up, because STI is the same metric used in a completely different context with the opposite goal.

In an open-plan office measured against ISO 3382-3, the design intent is for STI to fall quickly with distance — that is exactly what distraction distance and speech privacy are built on. You want a colleague’s phone call to become unintelligible a few metres away, because that unintelligibility is what gives everyone else privacy and lets them concentrate.

Inside a meeting room, the goal flips entirely. Everyone in the room is a deliberate participant in the conversation, so you want STI to stay high everywhere in the room — at the far end of the table, near the door, at the credenza where the laptop and speakerphone sit. There is no “privacy distance” you are trying to create within the four walls of a booked meeting room; there is only the requirement that every attendee, and increasingly every remote participant, hears every word cleanly.

Same metric, same physics, opposite design target. That distinction is the single most useful thing to understand about STI before you start treating a room.

What actually drives speech intelligibility in a room?

STI is not one thing — it is the combined product of several factors, each of which we check separately during a survey.

1. Reverberation time (RT60)

Reverberation time is the single biggest lever on STI in most meeting rooms. As sound reflects repeatedly off hard walls, glass, and ceilings, each syllable’s decay tail overlaps with the next syllable, smearing the modulation pattern that carries intelligibility. The target for meeting and conference rooms is RT60 of roughly 0.5–0.8 seconds — short enough that consonants stay crisp, long enough that the room doesn’t feel acoustically “dead” or unnatural.

Push RT60 much beyond 0.8 s in a meeting room and STI drops measurably, because reflections increasingly dominate the direct sound reaching each listener’s ear.

2. Background noise and signal-to-noise ratio

STI is a ratio problem as much as a reverberation problem. HVAC hiss, hallway noise bleeding through a thin partition, a projector fan, or an open-plan area adjacent to the room all raise the noise floor. Even in a room with well-controlled reverberation, a poor signal-to-noise ratio (loud background against a quiet talker) will drag STI down, because the modulation of speech gets partially masked by the modulation-free “static” of steady noise.

3. Room geometry and distance

Long, narrow rooms, rooms with hard parallel walls (flutter echo), and rooms where the talker and listener are far apart with no direct line of sight all reduce the direct-to-reverberant sound ratio reaching the ear — worsening STI even if the room’s overall RT60 number looks acceptable on paper.

4. Sound system and microphone placement (for hybrid rooms)

In rooms with ceiling speakers, soundbars, or paging systems, the electro-acoustic path itself has its own STI contribution — a poorly aimed or underpowered speaker, or a mic placed too far from typical talkers, adds its own smearing on top of the room’s native acoustic behaviour.

How glass-walled meeting boxes and hard finishes wreck STI

The modern glass-box meeting room — full-height glass partitions, a glossy table, a hard ceiling tile chosen for looks rather than performance, polished or laminate flooring — is close to a worst-case scenario for STI. Glass and hard gypsum board reflect the vast majority of speech-frequency sound rather than absorbing it, so RT60 balloons well past the 0.5–0.8 s target, sometimes into the 1.2–1.8 s range in small, hard rooms.

We see this constantly in conference room surveys: a beautifully designed glass meeting room that photographs well for the office tour but that nobody wants to sit in for a 45-minute client call, because by minute ten everyone is straining to follow the discussion. It is also, not coincidentally, the same failure mode we describe in why conference rooms turn into muddy meetings — long RT60, low STI, and fatigue are three symptoms of the same root cause.

The fix is not to remove the glass (which usually serves a legitimate transparency or branding purpose, and doesn’t affect sound transmission between rooms anyway — glass is really a sound insulation question, not an absorption one). The fix is to add absorption elsewhere in the room — ceiling, one or two walls, sometimes acoustic elements integrated into the table or credenza — to bring RT60 back down into the target band and restore a healthy direct-to-reverberant ratio. This is the same NRC-driven logic we use in every absorption-led project: identify the reflective surfaces, calculate the absorption needed via the Sabine relationship, and apply enough of the right material in the right place — usually ceiling treatment first, because the ceiling is almost always the largest unobstructed reflective surface in a meeting room.

How is STI actually measured?

STI is measured with a dedicated acoustic analyzer, not estimated by ear. The standard method, per IEC 60268-16, works roughly like this:

1. A loudspeaker positioned where a talker’s mouth would typically be plays a specific modulated test signal (a version of this is often called MLS — Modulation Transfer Function measurement — or the simplified STIPA signal used for field work).
2. A calibrated measurement microphone is positioned at one or more listener locations around the room — at the table, near the door, at the credenza where a speakerphone usually sits.
3. The analyzer compares the modulation depth of the signal that arrives at the microphone against the original signal across a set of frequency bands and modulation frequencies.
4. The degree of “modulation loss” — caused by reverberation and background noise — is converted into the 0–1 STI value using a standardized weighting (the Modulation Transfer Function, or MTF).
5. Background noise is measured separately and folded into the calculation, since noise masks the modulation just as reverberation does.

The field-friendly version of this process, STIPA, takes only a few minutes per position and is what we typically run during an on-site acoustic measurement alongside RT60 and background noise readings — because in a meeting room these three numbers tell almost the whole story together. If you’ve ever been handed an acoustic report and weren’t sure what the STI line meant, our companion guide on reading an acoustic report walks through exactly how RT60, STC, NRC, and STI relate to each other.

STI, RT60, and background noise — how the numbers connect

It helps to see the three metrics side by side rather than as separate topics, because in a real room they move together:

A room can have an “acceptable” RT60 on paper and still deliver disappointing STI if the background noise is high, or if the geometry puts listeners far from the direct sound path. That’s why we always measure STI directly rather than inferring it purely from an RT60 reading — it’s the number that actually reflects what people experience.

Why STI matters even more for hybrid and video-conferencing meetings

Reverberation and noise don’t just affect the people physically in the room — they affect what remote participants hear, often more severely. A conference-room microphone (whether it’s a tabletop unit, a ceiling array, or a laptop mic) captures the direct voice and every reflection and every background sound in the room, then sends that combined signal down the call. The remote listener has no way to mentally filter out the room’s reflections the way an in-room listener partially can — so a room with mediocre STI for in-person attendees can sound noticeably worse, muddier, and more fatiguing to someone dialing in.

This is one of the most under-appreciated findings from our project work: improving room acoustics measurably improves perceived call quality for remote participants, often as much as upgrading the microphone hardware itself — a topic we cover in more depth in office speech privacy and video calls. If your organization has invested in a premium video-conferencing camera and microphone bar but meetings still sound echoey to remote staff, the room’s acoustics — not the hardware — is very often the actual bottleneck.

Where STI fits alongside other meeting-room and office metrics

Speech intelligibility doesn’t exist in isolation — it’s one piece of the broader acoustic picture we assess for any office acoustics or conference-room project:

• Reverberation time (RT60) — covered above, and in detail in what is reverberation time.
• NRC and absorption coefficients — determine how much material you need to hit the target RT60; see NRC vs. sound absorption coefficient.
• Call-centre acoustics — a related but distinct challenge, where the goal is intelligibility on each individual call amid dozens of simultaneous conversations; see call center acoustics and the service page for call center acoustics solutions.
• Speech privacy in open-plan areas — the deliberately low-STI counterpart to this article; see distraction distance and speech privacy.

How we fix a meeting room with poor STI

In practice, our approach to a low-STI meeting room follows the same site survey → design → production → installation sequence we use across all room acoustics work:

1. Measure first. We take RT60, background noise, and STIPA readings at several listening positions — never guess based on how a room “feels” or looks.
2. Identify the dominant reflective surfaces. In most glass-and-gypsum meeting rooms, the ceiling and one or two hard walls are doing the majority of the damage.
3. Calculate the absorption required. Using the Sabine relationship (RT60 = 0.161 × V / A), we size the ceiling and wall treatment needed to bring the room into the 0.5–0.8 s target band.
4. Treat the ceiling first, then walls as needed. Ceiling absorption is usually the highest-impact, least disruptive intervention in a meeting room, since it doesn’t compete with whiteboards, screens, or glass walls.
5. Address background noise separately. If HVAC or adjacent-room noise is a factor, no amount of absorption will fully compensate — that’s a noise-source or sound insulation issue, and we flag it distinctly rather than trying to solve it with room absorption alone.
6. Re-measure. We confirm the post-treatment STI and RT60 against the target before calling the project complete.

This sequencing matters: absorption fixes reverberation-driven STI loss, but it will not fix a genuinely noisy HVAC system or a partition that’s letting hallway conversation bleed straight in. Setting realistic expectations up front — what absorption can and can’t do — is central to how we scope every conference room acoustics project.

Frequently Asked Questions

What is a good STI value for a meeting room?

Aim for 0.60 or higher, with 0.75+ considered excellent — appropriate for boardrooms, client-facing rooms, and rooms used heavily for video conferencing. Below 0.60, expect noticeably more repeated sentences and listener fatigue.

What is the difference between STI and reverberation time (RT60)?

RT60 measures how long sound persists in a room after the source stops; STI is a predictive intelligibility score that factors in RT60 along with background noise and signal-to-noise ratio. RT60 is one input to STI, not a substitute for it — two rooms with similar RT60 can have different STI if their noise floors differ.

Why do glass-walled meeting rooms often have poor speech intelligibility?

Glass, along with hard gypsum walls, glossy tables, and hard ceilings, reflects rather than absorbs speech-frequency sound. This pushes reverberation time well past the 0.5–0.8 s target for meeting rooms, smearing the modulation pattern of speech and lowering STI — even though the glass itself has nothing to do with sound passing between rooms, which is a separate sound-insulation question.

Does STI affect video-conferencing call quality?

Yes, often more than people expect. A room microphone captures both direct speech and every reflection and background noise in the room, and remote participants can’t filter that out the way in-room listeners partially can. Improving room acoustics frequently improves perceived call quality as much as upgrading microphone hardware.

Is a high STI always the goal in every space?

No — this is the key nuance. In meeting rooms and conference rooms, you want STI high everywhere in the room. In open-plan offices, the design goal for ISO 3382-3 is the opposite: STI should fall below 0.50 within a short “distraction distance” so that nearby conversations don’t intelligibly carry, giving occupants speech privacy.

How is STI measured in an existing meeting room?

A calibrated analyzer plays a standardized modulated test signal (commonly the STIPA method) from a loudspeaker at a typical talker position, while a measurement microphone captures it at one or more listener positions. The loss of modulation depth between source and receiver, combined with the measured background noise, is converted into the 0–1 STI score — typically measured alongside RT60 during a full acoustic measurement survey.

Ready to find out where your meeting rooms actually stand?

If your team is repeating themselves in the boardroom, or remote colleagues keep saying “you’re cutting out” on video calls, the room’s acoustics — not the technology — is the most likely culprit. Han Acoustic runs full site surveys covering RT60, background noise, and STI, then designs, produces, and installs the absorption treatment needed to bring your meeting rooms up to target. Contact us to schedule an on-site acoustic measurement for your conference rooms.