TLDR
Speech intelligibility is the degree to which speech can be understood by listeners in a given acoustic environment. It is not about volume — a room can be loud yet unintelligible if reverberation, background noise, or poor speaker placement smear consonants beyond recognition. The primary objective metric is the Speech Transmission Index (STI), standardised in IEC 60268-16:2020, which ranges from 0.00 (completely unintelligible) to 1.00 (perfect clarity). An STI of 0.60 or above is considered "good" for most speech applications. Speech intelligibility depends on three interacting factors: the direct-to-reverberant ratio (controlled by room absorption and distance), background noise level (controlled by HVAC and insulation), and the speech signal itself (level, directivity, frequency content). Designing for intelligibility is non-negotiable in classrooms, courtrooms, transport hubs, and emergency announcement systems.
Real-World Analogy
You are in a busy restaurant trying to follow your friend's story. She is speaking at normal volume, but the hard tile floors, glass walls, and fifty other conversations create a wall of reverberant noise. You catch maybe one word in three. Now imagine the same restaurant with upholstered booths, acoustic ceiling tiles, and carpet — suddenly every word is crystal clear. The difference is not that your friend spoke louder or that you listened harder. The room changed. Speech intelligibility is the measure of that change: how much of the speech signal survives the journey from mouth to ear through the acoustic gauntlet of the room.
Technical Definition
Speech intelligibility is formally quantified through the Speech Transmission Index (STI) per IEC 60268-16:2020. STI evaluates how well the temporal modulations of speech (the rapid envelope fluctuations that carry consonant information) are preserved as sound travels from source to receiver through a room.
The STI Scale
| STI Range | Rating | Typical Application |
|---|---|---|
| 0.00 – 0.30 | Bad | Unusable for speech |
| 0.30 – 0.45 | Poor | Significant listener effort required |
| 0.45 – 0.60 | Fair | Acceptable for casual conversation |
| 0.60 – 0.75 | Good | Classrooms, meeting rooms, courtrooms |
| 0.75 – 1.00 | Excellent | Recording studios, hearing-critical spaces |
Three Pillars of Intelligibility
- Signal-to-noise ratio (SNR): The difference between the speech level and the background noise level at the listener's ear. Every 1 dB improvement in SNR yields roughly a 3% improvement in word recognition scores up to saturation. Background noise from HVAC, traffic, and adjacent spaces directly degrades SNR.
- Reverberation: Late reverberation (energy arriving after ~50 ms) fills the gaps between syllables, masking the consonant transients that carry most of the linguistic information. The relationship between RT60 and STI is approximately inverse: doubling RT60 reduces STI by 0.10-0.15 points in a typical room.
- Direct-to-reverberant ratio: Determined by the listener's distance from the source relative to the critical distance. Listeners beyond critical distance receive more reverberant energy than direct energy, and STI drops accordingly.
Beyond STI
Other intelligibility metrics exist for specific contexts:
- STIPA (IEC 60268-16): A simplified, faster measurement version of STI using fewer modulation frequencies
- %ALcons (Articulation Loss of Consonants): An older metric that estimates the percentage of consonants lost, where values below 10% are considered acceptable
- D50 (ISO 3382-1): Definition — the ratio of early energy (first 50 ms) to total energy, which correlates strongly with speech intelligibility
Why It Matters for Design
- Education: ANSI S12.60 requires classrooms to achieve conditions supporting an STI of 0.60 or better. Research shows that a 0.10 reduction in STI correlates with a 10-15% drop in test comprehension scores for children, and the effect is even more severe for non-native speakers and children with hearing impairments.
- Emergency communication: Building codes and fire safety standards (BS 5839-8, NFPA 72) require voice alarm systems to achieve STI ≥ 0.50 throughout evacuation routes. Lives depend on passengers understanding announcements in airports, train stations, and shopping centres.
- Healthcare: In hospitals, misheard medication names or dosage instructions can be fatal. Acoustic design for nursing stations, consultation rooms, and operating theatres prioritises speech intelligibility alongside noise control.
- Legal proceedings: Courtrooms must ensure that witnesses, judges, and jury members can follow testimony without ambiguity. Many jurisdictions specify minimum intelligibility criteria for courtroom audio systems.
- Workplace productivity: Studies in open-plan offices show that irrelevant speech at STI > 0.50 is the primary cause of cognitive distraction. Paradoxically, the goal in open-plan design is to reduce intelligibility of overheard speech (using masking and absorption) while maintaining it for intended conversations.
How AcousPlan Uses This
AcousPlan computes STI for every simulation using the modulation transfer function method defined in IEC 60268-16. The results dashboard colour-codes STI on a green-amber-red scale and shows how it varies with listener distance from the source. When STI falls below your target, the AI diagnostic identifies the dominant cause — excessive reverberation, high background noise, or insufficient direct sound — and the auto-solve engine recommends targeted treatments. The speech privacy calculator separately evaluates intelligibility reduction for open-plan scenarios per ISO 3382-3.
Related Concepts
- What is STI? — The technical deep-dive into the Speech Transmission Index calculation
- What is Critical Distance? — The distance threshold that determines direct-to-reverberant ratio
- What Are Early Reflections? — The reflection energy that reinforces or degrades speech clarity
- What is the Lombard Effect? — How speakers unconsciously raise their voice in noisy rooms
- What is Acoustic Privacy? — The opposite goal: reducing intelligibility for confidentiality
Calculate Now
Check whether your room design supports clear speech — run a simulation in AcousPlan and see the STI score for every listener position, with actionable recommendations when it falls short.