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.

SiteCalc’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, Siri needs internet. On a job site with no cell signal, Siri doesn’t work. SiteCalc’s voice parser runs on-device. No connection needed.

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

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), voice is 3-4x faster than typing. Measure, speak, measure, speak. No phone fumbling 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. This one surprised me. I expected noise to kill voice input. Compressors, saws, radio, other crews yelling. But the phone’s microphone sits inches from your mouth when you’re talking into it, and modern noise cancellation in iOS handles the background reasonably well. Not perfect. In a room with an active chop saw, you’ll need to step a few feet away. But general job site noise doesn’t kill 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

I’ll be honest about the limits.

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 SiteCalc’s parser kicks in. If Siri routinely misunderstands you, SiteCalc’s voice input will have the same upstream problem. The dimensional parsing is solid once the words are transcribed correctly.

Background music with lyrics sometimes gets picked up as dimension words. “Let’s Go” by the Red Hot Chili Peppers is fine. A country song saying “five miles” in the background can confuse things if the timing is bad. Turn down the radio or get closer to the mic.

The Speed Difference

I timed it. Entering “12 feet 6 and 3/4 inches” by tapping on the screen takes about 8 seconds including finding the fraction input. Speaking it takes 3 seconds.

Over 50 entries in an estimating session, that’s 250 seconds saved. Four minutes. Not life-changing for one session, but across a week of estimating, across a career, it adds up. And the real value isn’t the seconds. It’s that you never have to put down what you’re holding.

SiteCalc includes voice input across all 85 calculators. Every dimension field accepts spoken input. Combine it with Siri Shortcuts to open any calculator hands-free. It’s not a separate mode or a premium feature. It’s just how you enter numbers when your hands are busy.

---