Audio Form
Some people will never type your form. They will talk to it.
A driver at a fuel stop. A nurse between rooms. Someone whose thumbs are not the fastest way they express themselves. An audio form asks the question out loud and listens to the answer.
What is an audio form?
An audio form is a form that is answered by speaking instead of typing. YourForm narrates each question with AI speech, transcribes the spoken reply, detects the language automatically, and writes the result into your Google Sheet — so respondents never touch a keyboard.
The problem
What is actually going wrong with your mobile form?
Open your own form on a phone and type a paragraph into it. The keyboard eats half the screen. The field you are filling scrolls out of view. Autocorrect rewrites a surname. You give up somewhere around the third open-ended question — and you are the person who built it.
Now picture someone doing that one-handed, outdoors, on a cracked screen, in a language your form was not written in. Typing is not a neutral interface. It is a filter, and it quietly removes the people whose answers you most needed.
The usual fix is to cut questions until the form is short enough to survive. That works, and it also means you stop asking the things you actually wanted to know.
Every open-text field on a mobile form is a small toll. Some people pay it. The rest close the tab.
The difference
Why not just add a voice-note field?
Plenty of tools let someone attach an audio recording. That solves the respondent’s problem and hands you a new one: a folder of clips nobody has time to listen to. The difference is what happens after the person stops talking.
| Voice-note upload | YourForm Audio Form | |
|---|---|---|
| What you receive | An audio file per respondent, to be played back one at a time | Structured text in the right column of your sheet, ready to filter and count |
| Language | Whatever was spoken; you work out which language later | Detected automatically — Nova-3 multilingual, no language picker to get wrong |
| Mistakes | Re-record the whole answer | “No, my name is Namo” updates that one field and moves on |
| The questions | Read silently off the screen | Spoken aloud, so the screen is optional |
| Follow-ups | None — it is a recording, not a conversation | Asks again when an answer is unclear, and can explain why a field is needed |
How it works
How an audio form works
- 1
It asks
Each question is spoken with ElevenLabs Flash v2.5, which is fast enough that the pause between question and answer feels like a conversation rather than a page load.
- 2
They answer
Speech streams to Deepgram Nova-3 in multilingual mode. There is no “select your language” step, because asking someone to declare their language before they have said a word is its own barrier.
- 3
It keeps up
Answers are matched to fields as the person talks. Correct yourself halfway through and the correction lands on the field you meant, not the one you are currently on.
- 4
It confirms
Before submitting, the form reads back what it captured. The respondent hears their own answers and can fix any of them by saying so.
Capabilities
What you get
Multilingual without configuration
Nova-3’s multilingual model handles code-switching mid-sentence — the way people actually speak when the technical noun is in English and the rest of the sentence is not.
Corrections that land in the right field
The hard part of voice input is not hearing the words, it is knowing which answer someone is revising. Say “actually, make that Tuesday” and the date field changes, not the last thing you were asked.
Answers to “why do you need this?”
People hesitate before giving a phone number. The form can say why it is asking, in one line, and then get on with it.
Your Google Sheet is the database
Answers land in a sheet you own, in real time. There is no separate dashboard holding your respondents’ data hostage, no export button to remember, and nothing to migrate if you ever leave.
Conditional logic, on the free plan
Show a question only when an earlier answer calls for it. Most builders put branching behind the paid tier; here it is included from the first form you make.
Change format without rebuilding
Questions, logic and responses are stored independently of how the form looks. Pick a different format from the sidebar and the same form is re-rendered — no re-typing, no lost answers.
Honest fit
Is an audio form right for what you are building?
Every format is wrong for something. Knowing which one you are in saves more time than any feature list, so here is where this one wins and where it does not.
Reach for it when
- Field teams, drivers and anyone whose hands are busy
- Respondents with low typing confidence or limited literacy
- Multilingual audiences you cannot segment in advance
- Long open-ended answers people would never type
- Accessibility requirements around motor and vision impairment
Pick another format when
- Loud environments — a factory floor will fight the microphone
- Long strings of exact characters, like a serial number or an IBAN
- Respondents who cannot grant microphone permission on a managed device
- Anywhere speaking answers aloud would leak something private
Questions
Audio Form questions, answered
- What is an audio form?
- An audio form is a form answered by voice. The questions are read aloud with AI speech, the respondent speaks their answers, and those answers are transcribed and written into your Google Sheet as normal text you can sort and filter.
- Does an audio form work in languages other than English?
- Yes. Speech recognition runs on Deepgram Nova-3 in multilingual mode, so the language is detected from what the person says rather than chosen from a dropdown beforehand. Mixing two languages in one sentence is handled too.
- Can someone correct an answer they already gave?
- Yes, by saying so. If the form heard “Nemo” and the respondent says “no, my name is Namo”, that one field updates while the rest of the conversation carries on. No restarting and no going back through earlier questions.
- Do I need my own ElevenLabs API key?
- Yes. Audio forms speak using your ElevenLabs account, which you connect once in the builder under the Audio Form tab. Keeping the key yours means the voice, the usage and the billing stay under your control.
- Where do the transcribed answers end up?
- In the Google Sheet connected to the form, in the column for that question, in real time. Columns are keyed by question ID and appended, so editing or reordering questions later never shifts data already collected.
- Is an audio form accessible?
- For many people it is more accessible than a typed form, particularly for motor impairments and low vision, since the questions are spoken and no keyboard is required. It is not a replacement for everyone — respondents who are deaf or hard of hearing are better served by another format, which is one reason the same form can be published in five.
The other four
The same questions, published four other ways
Format is a setting, not a rebuild. Your questions, logic and collected responses stay exactly where they are when you switch.
Build one and send it to five people.
That is usually enough to know whether an audio form suits your audience. Two forms and 200 responses a month are free, and responses go to your own Google Sheet from the first one.