Caption: “A short customer-service greeting (on this sample page, a tone stands in for it)”
Question 2: Listen to the clip. Is this voice a real person or AI-generated?
A real person41% · 409 people
AI-generated45% · 450 people
Not sure14% · 141 people
Q3 · 1–5 scale · 1,000 answers
How sure are you?
Question 3: How sure are you?
3.3
average out of 5
9% chose 1 (91)
16% chose 2 (160)
27% chose 3 (272)
29% chose 4 (290)
19% chose 5 (187)
1
2
3
4
5
1 = Just guessing5 = Completely sure
Q4 · Multiple choice · 1,000 answers
What made you decide? Pick all that apply.
Question 4: What made you decide? Pick all that apply.
The breathing and pauses32% · 323 people
The tone or emotion48% · 483 people
How words were pronounced20% · 199 people
The rhythm or pacing27% · 274 people
Background sound13% · 134 people
Nothing specific, just a feeling22% · 217 people
People could pick more than one, so shares add up to more than 100%.
Q5 · Short text · 1,000 answers
In a sentence, what did you notice about the voice?
The breathing sounded added in, always at the same spot. I’d still have picked up the call, though.
Age 61 (55–64) · Man · Nebraska (Midwest)
It never tripped over a single word, and real people do.
Age 23 (18–24) · Man · Idaho (West)
Every pause was exactly the same length. If I wasn’t listening for it, I might have missed it.
Age 23 (18–24) · Woman · Florida (South)
Honestly, I couldn’t tell. Parts sounded real and parts didn’t. I’d want to hear a longer clip.
Age 19 (18–24) · Man · Ohio (Midwest)
It was friendly, but the friendliness never changed. It took me two listens.
Age 64 (55–64) · Man · New Jersey (Northeast)
There was a small breath before the second sentence, which felt human.
Age 22 (18–24) · Woman · Pennsylvania (Northeast)
It hesitated on one word, the way people do. But I wouldn’t bet much on it.
Age 67 (65+) · Woman · Colorado (West)
It sounded real, but these days you can’t be sure.
Age 44 (35–44) · Man · California (West)
The export has every answer.
Recording studies
What a recording study returns
People answer in their own words on video or audio, or read your text aloud. 10 seconds to 2 minutes. You get each recording with who made it and a written transcript, for $25 per response.
Q1 · Recording · 100 answers
When did you last hear a voice you weren’t sure was real? Tell us what happened.
Age 31 (25–34) · Woman · Ohio (Midwest)
Video · 0:34· 4.9 MB· 004745f6
Sample still
Transcript: A couple of weeks ago I got a voicemail that sounded exactly like my bank, same greeting and everything. Then it asked me to read back a code, and my bank never does that. I hung up and called the number on my card instead.
Age 46 (45–54) · Man · Georgia (South)
Video · 0:41· 6.0 MB· 00c43329
Sample still
Transcript: It was a video of a news anchor someone shared online. The voice was right, but the lips were a little behind the words. I watched it three times before I decided it wasn’t real.
Age 68 (65+) · Woman · Arizona (West)
Video · 0:27· 3.8 MB· 038dfdd0
Sample still
Transcript: My grandson played me two clips and asked which one was him. I picked the wrong one. The fake even had the little laugh he does.
3 of 100 recordings shown. In your results, Play opens each one right here and Download saves the file; the export lists each recording’s file name and transcript.
Run a study like this one
This survey costs $1,750: 1,000 responses at $1.75 per complete. Recording studies are $25 per response.
Minimum 100 responses per study (20 for video). US adults.