Native - Speakers Recording Quality Assurance

Lifted (an Upwork Company); United States; Remote; USD 21,000 - 34,000; Full time

Job summary

Lifted (an Upwork Company) is seeking native Persian speakers for Recording Quality Assurance (RecQA) tasks. You will listen to existing audio, compare with prompt text, and fix minor text-audio mismatches for high-quality AI data.

The work is judgment-heavy, requires careful attention to detail, and is remote with approximately 7 hours per week. Payment is output-based, based on the number of correctly reviewed utterances.

Qualifications

• Native Persian speaker with strong listening and transcription skills.
• Experience with QA, audio review, or transcription preferred.
• Strong attention to detail and ability to identify subtle audio-text mismatches.
• Excellent listening comprehension for the assigned locale.
• Ability to apply defined issue categories consistently.
• Reliable availability for approximately 7 hours per week.

Responsibilities

• Listen to audio recordings and compare them with the corresponding prompt text.
• Identify and correct mismatches, mispronunciations, and punctuation where appropriate.
• Apply issue tags for major discrepancies (e.g., Mismatch, Noise).
• Maintain consistent review quality across all tasks and adhere to project guidelines.
• Complete tasks within project deadlines and communicate any issues clearly.

Skills

Native Persian speaker

QA experience

Audio QA

Job description

• Listen carefully to each audio recording and compare it against the corresponding prompt text.
• Identify subtle differences between the spoken audio and written text.
• Correct minor text errors, including missing or added words.
• Identify and correct mispronunciations where applicable.
• Adjust punctuation to accurately reflect pauses, questions, and other aspects of the spoken delivery.
• Apply the appropriate issue tag when the recording contains a significant mismatch or quality problem.
• Distinguish between minor text corrections and issues that require a defined quality tag.
• Follow project guidelines and apply judgment consistently across production units.
• Maintain careful, reliable, and consistent review quality throughout the project.

• Native Persian speaker
• Prior experience with QA, audio review, speech data, transcription, or similar quality-review work is preferred
• Strong attention to detail and ability to identify subtle audio-text mismatches.
• Excellent listening comprehension in the assigned locale.
• Ability to recognize issues such as missing words, added words, mispronunciations, unclear speech, and recording noise.
• Ability to apply defined issue categories accurately and consistently.
• Comfortable making careful judgment calls when reviewing audio recordings.
• Reliable availability for approximately 7 hours per week.

• Work anytime, provided you meet project deadlines and quality standards.
• The work is judgment-heavy, so contributors are expected to prioritize accuracy and consistency rather than simply maximizing speed.
• Tasks and overall workload may shift based on project needs and locale-specific audio volume.
• Consistent availability and high quality are required throughout the assignment.
• Payment is output-based, calculated according to the total number of correctly reviewed utterances.

Our enterprise client, a leading provider of high-quality, diverse datasets for AI model development is seeking skilled native speakers to support a Recording Quality Assurance (RecQA) initiative. In this role, you will listen to existing audio recordings, compare them against the corresponding prompt text, correct minor text-audio mismatches, and identify larger recording or quality issues using defined categories. Your careful review will help ensure high-quality speech data for the client's AI training initiative

Work Assignment and Task Summary (What to Expect)

Contributors will evaluate audio recordings against corresponding prompt text. The role requires carefully listening to each recording to make direct text edits for minor errors (e.g., missing/added words, mispronunciations, or punctuation) and applying specific issue tags (such as Mismatch, Cutoff, Mumble, or Noise) for major discrepancies, ensuring high accuracy and consistent quality across all reviews.

Originally posted on Himalayas