Audio Quality AI Trainer

@ Micro1

About the Audio Quality AI Trainer Role

Join micro1's team as an Audio Quality AI Trainer and contribute to the next generation of AI development. In this part-time, remote role (15-20 hours per week), you'll evaluate short audio clips, identify and classify audio artifacts, and provide precise annotations that directly impact model accuracy.

Key Responsibilities

• Analyze and classify audio artifacts such as noise, clipping, echoes, and compression issues.
• Filter and repair audio artifacts using spectral/manual edits and restoration workflows.
• Deliver accurate annotations to train and enhance AI audio models.
• Review and quality-assure annotations for consistency and correctness.
• Develop clear annotation notes for edge cases and ambiguous scenarios.
• Collaborate with the training team to refine project guidelines.
• Maintain sharp attention to detail during classification and QA processes.
• Provide constructive feedback to improve annotation quality and model performance.

Required Skills and Qualifications

• Experience in audio annotation, descriptive captioning, transcription, podcast/audio editing, or audio QA.
• Hands-on audio restoration using spectral repair, de-click/de-clip, denoise, de-reverb, EQ and manual fixes.
• Solid understanding of audio terminology and common recording faults.
• Proficiency with browser-based annotation tools and audio software (iZotope RX, Adobe Audition, REAPER, etc.).
• Ability to deliver before/after examples or quick demo edits.
• Exceptional auditory attention to detail and subtle audio discrepancy recognition.
• Strong written and verbal communication with clarity and precision.
• Commitment to annotation accuracy and quality assurance best practices.
• Availability to commit approximately 15-20 hours per week.

Preferred Qualifications

• Background in audio engineering, audio design, and audio restoration.

Key skills/competency

audio, AI, annotation, restoration, QA, classification, training, spectral, editing, engineering