STI — Speech Transmission Index
Speech Transmission Index
Related acoustics concepts: audio/room-acoustics and glossary/rt60.
STI is a standardized measure (IEC 60268-16) of speech intelligibility that produces a single value from 0 (completely unintelligible) to 1 (perfect intelligibility). It accounts for the two main physical degraders of intelligibility: reverberation (room reflections that smear speech in time) and background noise (which masks quieter speech sounds). STI is the primary metric used to specify and verify speech reinforcement systems, conferencing systems, evacuation alarm audio systems (life safety), and classroom acoustic designs.
STI Rating Scale
| STI Value | Intelligibility Rating | Typical Application |
|---|---|---|
| 0.75–1.00 | Excellent | Broadcast studio, critical listening |
| 0.60–0.75 | Good | Boardrooms, classrooms, conferencing |
| 0.45–0.60 | Fair | General office, lobby announcements |
| 0.30–0.45 | Poor | Reverberant spaces, noisy environments |
| 0.00–0.30 | Bad | Unusable for speech communication |
For most installed AV speech applications, STI ≥ 0.60 (Good) is the design target. NFPA 72 (fire alarm / mass notification systems) requires STI ≥ 0.70 in areas of refuge and other specific locations. IEC 60849 (sound systems for emergency purposes) uses similar thresholds.
What STI Measures
STI computes intelligibility by measuring the modulation transfer function — how well amplitude modulations in a speech signal survive transmission through the room and system. A room with high reverberation smears the amplitude envelope of speech, and a room with high background noise drowns out low-level modulations. STI quantifies both effects across seven octave bands (125 Hz to 8 kHz), weighted by their contribution to speech intelligibility.
The simplified version, STIPA (STI for Public Address systems), uses a specific two-channel test signal that can be measured in real time with a handheld analyzer (NTI Audio XL2, Rational Acoustics Smaart). STIPA is the practical field measurement method for installed AV systems.
Factors That Affect STI
Reverberation time (RT60): longer RT60 degrades STI. A room with RT60 > 1.0 second is difficult to achieve STI ≥ 0.60 without distributed speaker systems or absorptive treatments. See glossary/rt60.
Background noise: HVAC, mechanical equipment, traffic, and HVAC all raise the noise floor. Each 6 dB reduction in signal-to-noise ratio reduces STI by approximately 0.1 units. The target SNR for good intelligibility is ≥ 15 dB.
System distortion and frequency response: a speaker system with poor frequency response in the 1–4 kHz range (the most intelligibility-critical band) degrades STI even in an acoustically good room.
Echoes and discrete reflections: a strong echo (a reflection arriving more than 50 ms late with level within 10 dB of the direct sound) causes a large STI penalty disproportionate to its energy.
Measuring STI in the Field
STIPA measurement procedure:
- Play the STIPA test signal (from IEC 60268-16 or NTI Audio signal generator) through the loudspeaker system at normal operating level
- Place the measurement microphone at the listener position
- Measure with an STIPA-capable analyzer (NTI XL2, Rational Acoustics Smaart with STIPA plugin)
- Record the result at multiple seats across the coverage zone
- Report the minimum and average STI across measurement positions
For life-safety systems under NFPA 72, STI must be measured at 90% of measurement positions — 10% of positions below the threshold is acceptable.
Common Pitfalls
-
Designing for RT60 without checking STI. RT60 and STI are correlated but not the same metric. A room with acceptable RT60 (0.6 s) can still have poor STI if background noise is high or discrete echoes are present. Fix: always measure or predict STI directly; do not assume good RT60 implies good STI.
-
Measuring STI only at the front of the room. STI degrades toward the rear and sides of large rooms where speaker level is lower relative to the reverberant field. Fix: take measurements at a grid of positions across the listening area; report the worst-case and average values.
-
Confusing STIPA with STI. STIPA (simplified test signal method) and full STI (which requires more complex measurement) produce slightly different results. NFPA 72 and IEC 60849 specify which method is acceptable — verify the measurement method matches the standard being applied.
-
High STI measurement during day but poor intelligibility complaints. HVAC systems run at reduced capacity during quiet hours; during peak building occupancy, HVAC noise rises significantly and degrades STI. Fix: measure STI during actual operating conditions with full HVAC load, not during unoccupied hours.