AI Task Builder Collections lets you gather media files from Prolific's large, diverse participant pool, with annotation questions attached at the point of submission. Participants upload their files and answer your questions in one task, so every file arrives already labeled with the context you need.
How to set up a media collection
Follow these steps to build a media collection task using AI Task Builder.
Create your task
Go to AI Task Builder and select Create a new task
Select Data collection, then Continue
Enter a name for your collection, and then Continue
Add a task introduction explaining what participants need to do
Add task steps so participants know what to expect throughout
Set up your file upload
Add a page and select the file upload block
Add clear instructions for what participants should upload
Add helper text to specify accepted formats and size limits (for example: "Accepted formats: MP3, WAV, M4A. Max 100MB")
If a participant uploads a file in an unsupported format or exceeds the size limit, an error message will appear and they won't be able to continue until they upload a different fileSet the minimum number of files required (defaults to 1)
Set the maximum number of files allowed (defaults to 5)
Set the maximum file size (defaults to 10MB)
Select the file types you want to accept — defaults to JPG, JPEG, PNG, and PDF. See Which file types are supported? for formats and recommended size limits by media type
Add further content
Repeat the process to add more questions or content blocks, such as annotation questions, rich text, or images
Publish your study
Choose an existing project or create a new one, then select Set up study
Complete the study setup form as normal
Which file types are supported?
The file upload block accepts images, video, and audio. Use the tables below to choose the right formats and size limits for your study.
Images
Images default to JPG, JPEG, PNG, and PDF, you'll need to manually add any other formats you want to accept.
Use case | Accepted formats | Recommended max file size |
Standard photos (Android, web) | JPG, JPEG, PNG | 25MB |
Including iPhone photos | JPG, JPEG, PNG, HEIC, HEIF | 25MB |
High resolution or RAW | JPG, JPEG, PNG, TIFF | 50MB |
Note: HEIC and HEIF files won't preview in the file upload block, but will still be collected and included in your download.
Video
Use case | Accepted formats | Recommended max file size |
Short clips (under 2 min) | MP4, MOV, WEBM | 300MB |
Longer recordings (up to 5 min) | MP4, MOV | 500MB |
Extended recordings | MP4, MOV | 2000MB |
Note: the file size limit applies per file, not per submission. If you're collecting one recording per participant, set the maximum file count to 1.
Audio
Use case | Accepted formats | Recommended max file size |
Standard speech or voice | MP3, WAV, M4A, OGG | 100MB |
High-quality uncompressed audio | WAV, FLAC, AIFF | 500MB |
Note: iPhone Voice Memos saves as M4A by default. Include M4A in your accepted formats if you expect participants on iPhone.
The maximum file size you can set is 5GB. For most media tasks, the recommended limits in the tables above will give you the best results.
Tips for better media collection
Follow these tips to get consistently usable files and a smooth experience for your participants.
Give participants clear capture instructions
Add a rich text block at the top of your upload page to explain exactly how participants should capture their media - camera position, distance, duration, and what to include or avoid. Participants who understand what's needed produce far more usable data.
Be explicit about file formats and size limits
Don't rely on file type validation alone. Use the helper text field to spell out accepted formats and the size limit so participants know before they try to upload, for example: "Accepted formats: MP4, MOV. Max 300MB."
Keep one upload per page
One file upload per page, followed immediately by annotation questions, produces the cleanest and most consistently labeled data. If your task involves multiple recordings, use separate pages for each rather than increasing the maximum file count - this keeps annotations clearly paired with a single file.
Set your file size limit with headroom
Participants often record longer or higher-quality files than instructed. Add at least 50% headroom to your size limit to avoid unnecessary upload failures - for example, if you expect 30-second videos, set the limit high enough to accommodate 60-second recordings.
Get approval before collecting identifiable media
If your study involves video or audio where participants or others may be identifiable, you'll need prior written approval from Prolific before publishing. Collecting personally identifiable information without approval is a breach of our Researcher Terms. Learn more here: Can I ask participants for their personal information/identifiers?
How to download your dataset
Go to your study and select it by clicking the study name
Click Download task responses
This downloads a zip archive containing:
A JSON file with every participant's annotation responses
All uploaded media files, organised by collection, submission, and instruction ID
Every file you download is automatically named and organized so you can trace it back to the participant and question it came from.
When to export
You can export at any point during or after the study. Exporting early (after the first 10 to 20 submissions) is strongly recommended for media tasks; it lets you check file quality, format compliance, and annotation consistency before collecting the full sample.
If you need help setting up your study, contact our support team using the icon at the bottom right of this page.