GUIDE

What is a voice form, and when is it the wrong choice?

A voice form lets someone answer a website form by speaking instead of typing. Their speech becomes text, and software puts each answer in the right field for them to check before they submit. It helps when answers are long or awkward to type on a phone. It’s the wrong choice when the form is short, the details must be exact, or speaking isn’t private or possible.

By Tellfully · Updated October 4, 2026

Tellfully makes an optional voice assistant for website forms, so we have a stake in this topic. We’ve tried to be straight about where voice doesn’t fit.

How a voice form works

Most voice forms follow the same steps, whatever they’re called:

1. The visitor chooses to speak. The browser asks for microphone permission first.
2. Speech becomes text. A speech recognition service turns what they said into words.
3. Answers are matched to fields. Software works out which part of the answer belongs in which field: “the leak started after Tuesday’s storm” goes to the problem and to when it started.
4. Missing details are handled. Some voice forms fill what they can from one answer and leave the rest empty. Others ask a follow-up question about what’s missing.
5. The visitor reviews. They check the filled answers and fix anything that was misheard.
6. The visitor submits. The form sends the inquiry the way it always has, or the voice tool saves it, depending on the product.

Three kinds of voice form

Dictation into one field. Phones and computers already let people dictate into any text box. It’s free and familiar, but it fills one field at a time and doesn’t know what the form needs.

A one-sentence fill. The visitor speaks once and every field it can match fills together. It’s quick for forms where people know exactly what to say. TypelessForm works this way.

A short conversation. An assistant asks questions, listens and follows up on what’s missing. It takes longer, but suits forms where people need to explain something. Jotform AI Agents and Tellfully both work this way, in different settings: Jotform’s agents run on Jotform forms, and Tellfully works on the form already on your website.

There’s one more split worth knowing. Some voice tools replace your form with their own, and submissions go to their system. Others add to your existing form, so it keeps its fields, submit button and delivery.

When a voice form helps

• The answers are stories, not facts. “What’s going on with your roof?” is easier to say than to type.
• Visitors are on their phones. Typing a paragraph on a small keyboard is slow, and people tend to cut it short.
• People don’t know where their details go. A homeowner can describe the problem in one go and let the form sort out the fields.
• Your team keeps calling back for basics. A follow-up question about what’s missing can collect it before the inquiry arrives.

Whether it helps on your site is something to test, not assume. Compare submissions on your form before and after, over the same kind of period.

When a voice form is the wrong choice

• The form is short. Name, phone number and “call me back” take seconds to type. Adding voice there adds a step.
• The details must be exact. Speech recognition can mishear email addresses, phone numbers, street names and model numbers. If a form is mostly exact strings, typing is often better. Any voice form should show these on screen for the visitor to confirm.
• The information is sensitive. Don’t collect health details, payment card numbers or government ID numbers by voice. Some voice tools say outright that they aren’t built for this.^[4]
• It’s an emergency. Someone with water pouring through the ceiling needs a phone number, not a conversation with a form.
• Speaking isn’t private or practical. Open offices, job sites, waiting rooms and late-night browsing next to a sleeping partner all make typing the better option.
• Voice is the only option. A voice-only form shuts people out. More on that below.
• The tool can’t reach your form. Tools that add voice to an existing form need access to its fields. Forms embedded from another service in an iframe usually can’t be filled by a script on your page.

Accessibility: helpful for some, a barrier for others

Voice can make a form easier for people who find typing hard. People who can’t use a keyboard or mouse already use speech recognition software to get around the web.^[1]

But a form that relies on voice alone is a barrier. The W3C’s Web Accessibility Initiative lists “web-based services, including web applications, that rely on interaction using voice only” as a barrier for people with speech disabilities.^[2] People who are deaf or hard of hearing need to read what an assistant says, not just hear it.

So a voice form should:

• be optional, with typing available at every step and the original form still there;
• show what the assistant says as text, not only as audio;
• work with a keyboard, with visible focus;
• keep your form’s visible labels, so people using their own speech software can still say a field’s name to reach it.^[3]

Privacy: what to check

Before you add voice to a form, find out:

• Who processes the audio. Most voice forms send speech to an AI provider. Tellfully uses Google Gemini for live voice.
• Whether audio is recorded or kept. Tellfully doesn’t currently offer audio recording; transcripts and answers are separate from recordings.
• How long transcripts and answers are kept, and whether you control that.
• What visitors are told. They should know voice is processed by an AI service before they start, and that they can type instead.

Our privacy notice covers Tellfully’s details. For any other tool, read its privacy page and its AI provider’s terms.

The typing fallback isn’t optional

Microphones get blocked, permissions get denied and connections drop. When that happens, the visitor shouldn’t lose what they’ve already said or have to start over. A good voice form lets them switch to typing at any point, keeps their answers, and leaves the original form usable. If a voice form can only be finished by voice, someone will get stuck.

How to tell if it’s working

Measure submissions on your form, not conversations. A conversation that ends without a submission isn’t an inquiry. Note how many submissions you get over a few weeks before adding voice, then compare the same measure afterward. Read a sample of inquiries too: are they more useful to whoever calls back, or just longer?

Voice form terms, in plain English

Speech recognition (speech-to-text)

Software that turns spoken words into text. It can mishear names, numbers and unusual words.

Field mapping

Matching parts of an answer to the right form fields, such as putting “next Tuesday” in the preferred date.

One-sentence fill

A voice form that fills every field it can from a single spoken answer, without asking follow-up questions.

Conversational form

A form that asks questions one at a time, by text or voice, instead of showing every field at once.

Follow-up question

A question the assistant asks when an answer leaves out something the form needs.

Review step

The point where the visitor checks and corrects the filled answers before anything is sent.

Existing-form assistant

A voice or chat helper that fills the form already on your website, so the visitor still submits it there and its delivery stays the same.

Typing fallback

A way to continue by typing, without losing answers, whenever voice isn’t possible or wanted.