Voice Input for Construction Calculators

The Glove Problem

You’re on a roof in November. It’s 38 degrees. You’re wearing insulated gloves because your hands stopped working twenty minutes ago. You need to check a rafter length. Your phone is in your jacket pocket.

Take off gloves. Fish out phone. Unlock it. Open the calculator. Fat-finger 12 instead of 10 because your fingers are numb. Delete, retype. Get the answer. Put phone back. Put gloves on. Your hands are now colder than before.

This happens dozens of times a day on cold weather job sites. And it’s not just cold. Muddy hands in earthwork. Wet hands on a plumbing rough-in. Gloved hands running a saw. Paint-covered hands during finish work. There are a hundred situations where pulling out your phone and tapping a screen is the worst way to enter a dimension.

How Voice Input Should Work

Say the dimension. The app enters it. That’s the entire interaction.

Sounds simple but most voice-to-text systems aren’t built for construction input. Siri will transcribe “twelve feet six and three-quarter inches” as text. It won’t know that means 12’ 6-3/4” as a dimensional value. You’d still need to convert it manually.

Trussly’s voice parser is built specifically for construction dimensions. It understands:

Feet and inches together. “Fourteen feet eight inches” becomes 14’ 8”. “Seven foot three” becomes 7’ 3”. It handles both “feet” and “foot,” and you can skip saying “inches” at the end.

Fractions. “Three-quarter inch” becomes 3/4”. “Five and one-half inches” becomes 5-1/2”. “Nine sixteenths” becomes 9/16”. These are the fractions contractors actually use on tape measures, and the parser knows the common denominators: halves, quarters, eighths, sixteenths.

Mixed dimensions. “Twelve feet six and three-quarter inches” becomes 12’ 6-3/4”. The full feet-inches-fraction chain in one spoken phrase.

Bare numbers. “Twenty-four” becomes 24. Sometimes you just need a count or a bare measurement. No units forced on you.

Filler Word Stripping

People don’t talk in clean dimensional phrases. They say “uh, about twelve feet, um, six inches.” They say “let’s see, it’s fourteen foot, no wait, fourteen foot three.” They say “okay so the run is twenty-two feet.”

The parser strips filler words before processing: “uh,” “um,” “about,” “so,” “okay,” “let’s see,” “it’s,” “the.” What’s left is the dimensional content.

“Okay so it’s about twenty-two feet six” becomes 22’ 6”.

“Um, fourteen and a half inches” becomes 14-1/2”.

This matters because job site speech isn’t clean dictation. You’re talking while holding a tape measure, squinting at a number, maybe shouting over a compressor. The parser has to be tolerant of how people actually talk.

Why Not Just Use Siri?

Three reasons.

First, Siri transcribes to text. You still have to parse the dimension yourself and type it in. Voice-to-text isn’t voice-to-value.

Second, connectivity. Trussly asks iOS for on-device speech recognition where the phone supports it, so it can work without signal. On older devices, or before the on-device speech model has downloaded, Apple processes the audio on its servers and you need a connection.

Third, Siri doesn’t know construction. It’ll transcribe “two by four” as “2x4” sometimes, but it doesn’t know that “ten foot six and three-eighths” should become a single dimensional value of 10’ 6-3/8”. It’s a general-purpose transcription tool. Construction dimensions need a specialized parser.

Real Usage Patterns

Voice works especially well for the calculations that come up mid-task: concrete volume for a slab while you’re setting forms, stud counts while you’re walking a framed wall, or rebar spacing while you’re tying a mat. After watching how people actually use voice input on job sites, some patterns show up.

Rapid entry. When you’re measuring and entering a series of dimensions, like wall lengths around a room, you can measure, speak, measure, speak without fumbling with the keyboard between measurements.

One-handed operation. Tape measure in one hand, phone in the other. Tap the mic button with your thumb, say the number, done. Both hands stay occupied with useful work.

Noisy environments. Loud saws and compressors make any speech recognition worse. Holding the phone close helps. When it mishears, you’ll see it in the field before you use it.

The correction pattern. People make mistakes. They say the wrong number. A good voice input system needs to be easy to redo. Tap mic, say the right number, it replaces the wrong one. No need to delete characters and retype. Just re-speak it.

What Doesn’t Work Yet

Very complex compound dimensions are hit or miss. “Forty-seven feet, eleven and thirteen-sixteenths inches” works. But if you start throwing in multiple dimensions in one breath (“forty-seven feet eleven and thirteen-sixteenths by twenty-three feet four and a half”), the parser handles the first dimension and waits for you to enter the second one separately. One dimension per voice input.

Heavy accents can cause issues with the Apple speech recognition layer that runs before Trussly’s parser kicks in. If Siri routinely misunderstands you, Trussly’s voice input will have the same upstream problem. The dimensional parsing is solid once the words are transcribed correctly.

Background talking or music with spoken numbers can get picked up. Turn the radio down or get closer to the mic.

Check What It Heard

Speech recognition makes mistakes, especially with similar-sounding numbers like fifteen and fifty. Every recognized value lands in a field you can see before you calculate. Glance at it. It takes a second and it’s the difference between 15 and 50 cubic yards.

Trussly has voice input on 63 of its calculators. It’s not a separate mode or a paid add-on. It’s just how you enter numbers when your hands are busy.

---