How to Measure and Improve Speech Transmission Index

By Sarah Okonkwo · March 28, 2026

If you’ve ever mixed a live panel and watched the audience lean forward, or edited a podcast where the guest “sounds clear” but listeners still miss half the words, you’ve run into the same problem: speech intelligibility. It’s not just about volume or EQ polish—it’s about whether speech survives the room, the system, and the noise floor with enough clarity for the brain to decode it effortlessly.

That’s exactly what the Speech Transmission Index (STI) is designed to quantify. STI turns “Can you understand what they’re saying?” into a repeatable measurement you can track, compare, and improve. For audio engineers, musicians running backing tracks with announcements, podcasters treating home rooms, and studio owners tuning control rooms, STI helps you separate guesswork from results.

This guide breaks down what STI actually measures, how to test it in real spaces (from home studios to venues), how to interpret the numbers, and the practical fixes that move the needle—without turning your room into an anechoic chamber or over-processing voices into harshness.

What Speech Transmission Index (STI) Really Measures

STI in plain language

STI is a metric (0.0 to 1.0) that estimates how intelligible speech will be after it passes through a sound system and a room. It’s strongly influenced by:

Reverberation (reflections smearing consonants)
Background noise (HVAC, crowd, street noise, amp hiss)
Distortion and nonlinearity (overdriven amps, limiters pumping)
Frequency response (missing presence range, excessive low-mid buildup)
Time variance (AGC, noise reduction, gating, heavy compression)

Typical STI ratings (practical expectations)

0.00–0.30: Bad (words are frequently missed)
0.30–0.45: Poor (usable only in quiet conditions)
0.45–0.60: Fair (borderline for critical speech)
0.60–0.75: Good (solid intelligibility for most listeners)
0.75–1.00: Excellent (rare in reverberant or noisy spaces)

Real-world reference points:

Podcast voice in a treated room: 0.75+ is achievable
Classroom or meeting room with basic treatment: 0.60–0.75 is a good target
Small club with reflective surfaces and crowd noise: 0.45–0.65 is common, improving toward 0.60 is a win
Highly reverberant hall: may struggle to exceed 0.50 without distributed reinforcement and treatment

STI vs. STIPA vs. “can I hear it?”

You’ll often see STIPA (Speech Transmission Index for Public Address). STIPA is a standardized, faster-to-measure STI method widely used for PA systems and building compliance. In practical audio work, when someone says “STI measurement,” they often mean STIPA using a dedicated test signal and analyzer.

Also, subjective listening tests matter, but STI helps you:

Compare changes objectively (before/after treatment or DSP tweaks)
Quantify intelligibility across listener positions
Catch issues that sound “fine” near FOH but collapse at the back wall

When STI Matters Most (Real-World Scenarios)

Live events: Corporate talks, theater dialog, worship services, announcements between sets
Rehearsal rooms and studios: Producer talkback, booth communication, cue systems
Podcasting and voiceover: Home rooms with fan noise and reflections
Streaming and broadcast: Remote guests, IFB clarity, noisy locations
Installations: Gyms, airports, retail paging, classrooms

Tools and Equipment for Measuring STI

Dedicated STI/STIPA meters (fast, straightforward)

For engineers who need repeatability and speed, a dedicated instrument is the cleanest workflow. Features to look for:

STIPA-compliant measurement
Logging and reporting (CSV/PDF export)
Support for background noise correction
Octave band data (helps diagnose issues)
Calibrator compatibility (94 dB/114 dB, 1 kHz)

Measurement mic + audio interface + software (flexible, studio-friendly)

Many home studio owners and podcasters already have most of what’s needed:

Measurement microphone (calibrated omni preferred)
Audio interface with clean preamp and stable clock
Software that supports STI/STIPA or intelligibility-related analysis

This route is excellent for diagnosis, but be careful: not all acoustic measurement tools calculate STI correctly, and system processing (AGC, noise reduction) can skew results if not controlled.

Signal source and playback chain

To measure STI in a venue or room with reinforcement, you need:

A STIPA test signal (or STI-compatible modulated noise) played through the same path as speech
Consistent playback level (don’t “ride” the fader during measurements)
Access to system DSP settings (EQ, limiters, delays) so you can document changes

Step-by-Step: How to Measure STI (Practical Workflow)

Step 1: Define the goal and listening areas

Don’t measure one spot and call it done. Choose positions that represent real listeners:

Front row / nearfield
Middle seating area
Back of room
Off-axis seats (far left/right)
Balcony (if applicable)

For podcasts/home studios, pick:

Primary mic position (where the host sits)
Guest position (if different)
Standing position (if you record narration standing)

Step 2: Stabilize the system and the environment

STI is sensitive to changing conditions. Before measuring:

Disable any automatic gain control (camera inputs, conferencing software, stream encoders)
Avoid live noise reduction during the test if it will not be used for speech, or keep it enabled if it will be used all the time (just be consistent)
Set HVAC to a normal operating mode (don’t measure with HVAC off if it’s always on during shows)
Verify you’re measuring the actual speech chain (mic → preamp → DSP → amps → speakers)

Step 3: Calibrate and set a realistic speech level

STI depends on the signal-to-noise ratio. Set levels based on real use:

Place your measurement mic at ear height (roughly 1.2–1.7 m depending on seated/standing audience).
Calibrate the mic or meter if your workflow supports it.
Set a target level similar to speech:
- Meetings/classrooms: often around 60–70 dB SPL (A-weighted) at listener positions
- Live venues with music: speech may need to sit clearly above ambient noise; measure at realistic show conditions when possible

Tip: If you tune a PA with the speech channel far louder than anyone will tolerate, STI can look great on paper and fail in reality.

Step 4: Play the STI/STIPA test signal and capture measurements

Play the STIPA test signal through the system (ideally through the same input channel used for speech, or a dedicated test input that follows the same processing).
Measure at each listener position for the required averaging time (often 10–20 seconds depending on the method).
Log results and note:
- Location and distance
- System configuration (EQ presets, delays, limiter states)
- Ambient noise conditions (empty room vs occupied)

Step 5: Interpret the STI map, not just the best score

What you’re looking for is consistency. A room that reads 0.72 in the front and 0.42 in the back has a coverage and/or reverberation problem. A room that sits at 0.60–0.65 everywhere is often more usable than one with a single “excellent” hotspot.

How to Improve STI: Fixes That Actually Move the Number

1) Increase useful direct sound (coverage and speaker strategy)

STI improves when listeners get more direct speech relative to reflections and noise.

Re-aim speakers to keep energy off reflective walls and ceilings.
Use narrower directivity where appropriate (horns/waveguides) to control spill in reverberant spaces.
Consider distributed reinforcement (delay speakers) for long rooms instead of blasting the front fills.
Time-align delays carefully; misaligned delays can smear articulation and reduce intelligibility.

Live-event scenario: In a long, reflective hall, pushing the mains louder often increases reverberant buildup more than it increases clarity at the back. Properly delayed fills can raise STI in rear seats without punishing the front row.

2) Reduce reverberation with targeted acoustic treatment

You don’t need to treat every surface, but you do need to treat the right ones.

Ceiling cloud panels above audience areas can reduce early reflections that mask consonants.
Rear-wall absorption helps reduce slapback that destroys intelligibility for mid/back seats.
Carpet alone isn’t enough; speech clarity is often killed by mid/high reflections off drywall, glass, and ceilings.

Home studio/podcast scenario: If your mic technique is solid but speech still feels “phasey” or “roomy,” treat first reflection points near the desk and add absorption behind the speaker. STI improves when early reflections are controlled, not when you simply roll off highs.

3) Improve signal-to-noise ratio (SNR) the smart way

Noise is the enemy of STI. Attack it at the source:

Turn down HVAC where feasible; service rattling vents
Reduce stage noise bleeding into speech mics (amp isolation, drum shields where appropriate)
Use close mic technique and the right microphone pattern
Minimize system hiss: gain-stage properly, avoid excessive make-up gain into noisy components

Podcast scenario: A loud computer fan 1 meter away can ruin intelligibility more than a “less expensive” microphone ever would. Moving the computer, using quieter cooling, or adding isolation can raise STI more than swapping mics.

4) Optimize mic choice and placement for intelligibility

Dynamic cardioids are often forgiving and reject room noise well in untreated spaces.
Shotguns can work in controlled interiors, but reflections in small rooms can make them sound hollow; placement is critical.
Lavalier mics vary wildly with clothing noise and placement; consider headworn mics for consistent distance in live speech.

Key placement habits:

Keep the mic close (distance is your SNR)
Aim rejection nulls toward loud sources (stage wedges, drums, PA spill)
Use pop control and wind protection so you don’t over-EQ plosives later

5) Use EQ and dynamics to support speech, not fight the room

EQ can help STI indirectly by improving clarity, but it can’t fix a bad coverage plan or a highly reverberant room.

Practical speech EQ targets (starting points):

High-pass filter to reduce rumble and proximity boom (often 80–120 Hz depending on voice/mic)
Control muddy low-mids (often 200–400 Hz) if the room builds up
Add presence carefully (often 2–4 kHz) for articulation
Avoid harshness and sibilance (often 5–8 kHz) especially on bright condensers

Compression and limiting:

Moderate compression can help keep quieter words audible above noise
Over-compression can bring up room tone and reverberation, lowering perceived intelligibility
Avoid aggressive gates; chopped consonant tails reduce intelligibility fast

Common Mistakes to Avoid

Measuring in an unrealistic condition: empty room when it’s normally full, or HVAC off when it’s normally on.
Chasing only the highest STI number: a consistent “good” across seats beats an “excellent” near one speaker.
Using STI as a substitute for listening: measurements guide decisions, but always verify with real speech playback.
Forgetting system processing: conferencing DSP, stream encoders, and AGC can radically alter intelligibility compared to your local monitoring.
Trying to EQ away reverberation: cutting highs may reduce perceived splash, but it also removes consonant information.
Ignoring microphone technique: mic distance and orientation can make or break STI in noisy environments.

Quick Field Checklist: Boost STI Before the Show or Session

Walk the room with a speech track and listen for “blur” in the back
Confirm speaker aiming; keep energy off hard rear walls when possible
Set speech channel HPF and reduce low-mid buildup
Check gain staging to minimize hiss and avoid limiter pumping
Verify mic placement and pattern choice for the talker
Measure STI at 3–6 key listener positions and log results

FAQ

What’s a “good” STI for a podcast or voiceover?

For close-mic voiceover in a controlled home studio, 0.75+ is a realistic target. If you’re around 0.65–0.75, it can still sound professional—especially with good mic technique and low noise—just watch for reflections and fan noise.

Can I measure STI with a standard measurement mic and free software?

Sometimes, but not all tools implement STI/STIPA correctly. If intelligibility compliance or repeatable reporting matters (venues, installs, safety paging), a STIPA-capable meter or verified software workflow is worth it. For home tuning, you can still gain insight by measuring room acoustics (RT60, early reflections) and improving SNR, even if STI itself isn’t available.

Why did my STI drop after adding compression?

Compression can raise quiet words, but it can also raise reverb tails, crowd noise, and room tone. Heavy compression may make speech feel “louder” while becoming less intelligible. Try gentler ratios, slower release, or reducing makeup gain.

Does adding high frequencies always improve STI?

No. A presence lift (around 2–4 kHz) can help consonants, but too much top end can increase sibilance and listener fatigue. Also, if the room is reverberant, boosting highs can make reflections more obvious. Address reflections and coverage first.

What matters more for STI: loudness or clarity?

Clarity. Loudness helps only if it improves the signal-to-noise ratio at the listener position without increasing reverberant buildup or distortion. Past a point, louder can actually reduce intelligibility in reflective rooms.

How many measurement positions should I use in a venue?

For small rooms, 3–6 positions is a practical baseline (front/mid/back plus off-axis). For larger venues, measure more points and build a simple “STI map” so you can see coverage gaps and problem zones.

Actionable Next Steps

If you want better speech intelligibility you can trust, start simple:

Measure STI/STIPA at several listener positions (or measure related room metrics if STI tools aren’t available).
Fix coverage first: aiming, directivity, and delays for consistent direct sound.
Control reflections with targeted acoustic treatment (ceiling and rear wall are frequent winners).
Improve SNR with quieter sources, closer miking, and clean gain staging.
Refine EQ and dynamics to support articulation without raising the room.

If you want more practical tuning workflows, mic technique tips, and measurement-based studio upgrades, explore more guides on sonusgearflow.com.