Recorded data for AI teams
Voice and video of real people, recorded for your model.
Write the script people read aloud, or the questions they answer in their own words. Each person records in short parts on their phone or computer, and you get every file with a transcript and their age, gender and state and, for members who agreed, their city, when they took part, and their device and connection.
From a panel of 137,863+ adults who signed up to take part in paid studies.
Part 1 of 5 · Read aloud
“Turn the temperature down to sixty-eight, then find a coffee shop on my way that’s open right now.”
Age 41 (35–44) · Man · Ohio (Midwest)
Audio · 0:47· 4.3 MB· WAV · 48 kHz · processing off· k3V9q2Ma
Transcript: Turn the temperature down to sixty-eight, then find a coffee shop on my way that’s open right now.
Part 2 of 5 · Answer
How do you use a voice assistant in the car? Tell us in your own words.
Age 29 (25–34) · Woman · Texas (South)
Video · 1:44· 31.0 MB· Pw4nE8sY
Transcript: Mostly for directions and to call people while I drive. It gets street names wrong maybe one time in five, so I end up pulling over and typing it in, which is the thing I was trying not to do.
How it works
From your script to their recordings
You set up the session yourself. We recruit the people, take their consent, collect every recording and pay them.
- 1
Write the parts
Add up to 30 parts of up to 2 minutes each. In each one, people read your text aloud or answer your question in their own words, on video or as audio only.
- 2
Say how you’ll use it
State the use in plain words, such as training or evaluating a speech model. Each person reads it, word for word, before they agree to record.
- 3
Pick who takes part
Choose an age range, gender, and US regions or states. Before you launch, you see about how many people match.
- 4
We check it and send it out
A person at StudyScout reviews every study before it goes to the panel. Then we invite the members who match.
Who records
Real people who said yes to your use
US adults 18 and over, from a panel of 137,863+ adults who signed up to take part in paid studies.
Consent before anything records
Each person sees what will be recorded, who gets it and your stated use, and agrees first. They can stop any time before they submit, and then you get nothing.
Checked to be in the US
Before they start, each person confirms a US mobile number with a texted code, and VoIP numbers are refused. A session from a network outside the US is refused too.
Their best take of each part
People can record each part up to 3 times and send the one they’re happy with. A recording stops itself at the part’s longest length and can’t be sent shorter than its shortest.
What you get
Every file, its transcript and who recorded it
Your results page fills in as people finish: each recording with its transcript, and the person’s age, gender and state and, for members who agreed, their city, when they took part, and their device and connection.
Every recording
Each accepted recording, video with sound or audio only, to play on your results page or download.
A transcript of each
What each person said in each part, written out, with contact details masked.
Who said it
Each person’s age and age band, gender, US state and region, on every recording. Never their name, email or contact details.
A CSV of the whole study
A row per person, with each part’s file name and transcript, then their age, gender, state and region, plus, for members who agreed, their city, when they took part, and their device and connection.
For data teams
Formats and files for your pipeline
Uncompressed audio (WAV), no processing
Each recording is a WAV file made on our page with the browser’s echo cancellation, noise suppression and automatic volume off; people need Chrome or Edge on a computer, or an up-to-date iPhone or Android. Each one is listed with its sample rate and whether processing was off.
A script read line by line
A script of several lines is shown one line at a time while the part records one continuous take, and the CSV marks where each line is in the file.
Files by API
List a study’s recordings with their transcripts and who made them, get a signed download link for each file, and export the CSV, with your API key.
Pricing
Pay per person, by how long a session runs
Top up by card. Each price includes the participant’s pay, and you’re charged only for accepted recordings.
Up to 10 minutes
$10per person
20 people: $200
Up to 30 minutes
$25per person
20 people: $500
Up to an hour
$45per person
20 people: $900
A session’s length is its parts’ longest lengths added up. At least 20 people per study. Studies go to US adults 18 and over.
You may reject up to 10% of a study’s accepted responses, from its results page, while it runs and for 7 days after it closes; a rejected response isn’t charged.
FAQ
Questions about recorded data
- Can we use the recordings to train AI?
- Yes, when your study says so. You state the use in plain words, such as training or evaluating a model, and each person reads it before they agree to record. The recordings and their transcripts are licensed to you for the use you state.
- Do people read a script, or talk in their own words?
- Either, part by part, and one session can mix both. A script is shown on screen while they record, up to 1,500 characters a part. For an answer, you write the question and they reply in their own words.
- Video or audio?
- Either, part by part: video with sound, or audio only. An audio part can record uncompressed WAV with the browser’s echo cancellation, noise suppression and automatic volume off.
- How long can a session be?
- Up to an hour: up to 30 parts of up to 2 minutes each, each one recorded, sent and checked on its own.
- Who takes part?
- US adults 18 and over from a panel of 137,863+ adults who signed up to take part in paid studies, recording on their own phone or computer. You pick the age range, gender and US regions or states, and see about how many people match before you launch. Each one has verified a US mobile number. A session from a network outside the US is refused.
- What if a recording isn’t usable?
- You pay only for accepted recordings: responses screened out by our quality checks aren’t charged. You may reject up to 10% of a study’s accepted responses, from its results page, while it runs and for 7 days after it closes; a rejected response isn’t charged.
- Can I see recordings while it runs?
- Yes. Every study is reviewed before it goes to the panel. Once it’s live, each recording shows up on your results page as people finish, and its transcript follows.
Start your first recording session
Write your script or questions, state your use, pick who takes part, and get every file.
Recording sessions are $10 a person for up to 10 minutes, $25 for up to 30 minutes and $45 for up to an hour, at least 20 per study. The price includes each participant’s pay.