Skip to main content

Why quality matters

The data you contribute on WorkBolt is used to train AI models. The quality of your submissions directly impacts model performance. Maintaining high standards ensures better outcomes for clients and more opportunities for you.

Quality metrics

WorkBolt reviews submissions across six key quality dimensions:
MetricDescription
AccuracyHow closely your submission follows the script or instructions
CompletenessWhether all required elements are included
ConsistencyUniform pace, volume, and style across your recordings
Background noise levelAmount of ambient noise in your recordings
Voice clarityHow clear and intelligible your speech is
Microphone qualityTechnical quality of your recording equipment and setup
Each metric is scored during review, and the combined assessment determines whether your submission is accepted or rejected.

Submission review process

1

Submit your work

Complete the task and submit through the WorkBolt platform. Your submission status changes to Reviewing.
2

Quality review

A reviewer evaluates your submission against the project’s quality standards, scoring each quality metric.
3

Decision

Your submission is either Accepted (approved for payment) or Rejected (with feedback on what to improve).
4

Feedback

Rejected submissions include a reviewer’s note explaining the issues. Use this feedback to improve future submissions.

Audio quality standards

For voice recording projects, your audio must meet these requirements:
StandardRequirement
Background noiseMinimal — no audible traffic, appliances, pets, or people
Echo / reverbNone — use a treated or small, soft-furnished room
ClippingNo distortion from input levels that are too high
Signal-to-noise ratioSpeech should be clearly louder than any ambient noise
ConsistencyUniform volume and mic distance across recordings
Plosives / sibilanceMinimized — use a pop filter when possible
See Recording Setup for guidance on achieving these standards with your equipment.

Common reasons for rejection

Audible sounds in the background (traffic, fans, typing, other people). Record in a quiet, isolated space with doors and windows closed.
Not following the provided script or instructions exactly. Read back each line carefully before submitting.
Distortion caused by input levels that are too high. Reduce your microphone gain and maintain consistent distance from the mic.
Mumbling, speaking too quickly, or inconsistent volume. Speak clearly and at a natural, steady pace.
Missing recordings, skipped prompts, or partially completed tasks. Review all items before final submission.
Using a language, dialect, or accent that doesn’t match the project requirements. Only accept projects that match your profile.
Poor recording equipment producing muffled, tinny, or noisy audio. Consider upgrading your microphone — see Recording Setup.
Repeated rejections affect your eligibility for future projects. Review the feedback provided and improve your setup and process before submitting again.