Speech Audio Specialist - AI Trainer

Mercor; São Paulo; Sur place; BRL 90 000 - 120 000; Plein temps

Résumé du poste

Mercor is seeking an audio engineer in São Paulo to edit and quality-check Brazilian Portuguese voice recordings for speech and TTS applications. The role focuses on preparing high-quality datasets for ML pipelines, ensuring clarity, consistency, and technical compliance at scale.

Native or near-native Brazilian Portuguese fluency is required to assess pronunciation and naturalness. The ideal candidate will optimize audio quality, work with large batches, and collaborate with teams across

Qualifications

• Native or near-native Brazilian Portuguese speaker with listening intuition for speech nuances.
• Experience editing speech or voice recordings (not music production).
• Strong understanding of speech audio quality standards and common issues in dialogue.
• Experience working with high-volume audio datasets or structured workflows.

Responsabilités

• Edit Brazilian Portuguese voice recordings for use in TTS and speech AI systems.
• Perform audio cleanup including silence trimming, breath reduction, de-clicking, de-essing, and noise reduction.
• Conduct quality assurance checks for clipping, distortion, background noise, artifacts, and loudness.
• Ensure audio meets strict technical specifications for ML training pipelines.
• Evaluate recordings for pronunciation accuracy, naturalness, and fluency.
• Collaborate with teams working on Brazilian Portuguese datasets, talent recordings, and model training workflows.

Connaissances

Native Brazilian Portuguese

Speech/voice editing

Audio quality standards

Critical listening

High-volume datasets

Outils

iZotope RX

Pro Tools

Description du poste

OverviewWe're looking for audio engineers with strong experience editing and quality-checking voice recordings for speech and text-to-speech (TTS) applications, specifically for Brazilian Portuguese-language audio. This role focuses on preparing high-quality audio datasets for machine learning systems, ensuring clarity, consistency, and technical compliance across large volumes of Brazilian Portuguese speech data. Native or near-native Brazilian Portuguese fluency is required to accurately assess speech quality, pronunciation, and naturalness.

What You'll Do

• Edit and process Brazilian Portuguese voice recordings for use in TTS and speech AI systems

• Perform audio cleanup including silence trimming, breath reduction/removal, de-clicking, de-essing, and noise reduction

• Conduct detailed quality assurance checks for:

  • Clipping and distortion

  • Background noise and artifacts

  • Loudness consistency and level balancing

• Ensure audio meets strict technical specifications for ML training pipelines

• Evaluate recordings for Brazilian Portuguese pronunciation accuracy, naturalness, and fluency

• Work with large batches of recordings, maintaining consistency and throughput

• Collaborate with teams working on Brazilian Portuguese speech datasets, voice talent recordings, and model training workflows

Basic Qualifications

• Native or near-native Brazilian Portuguese speaker with strong listening intuition for Brazilian Portuguese speech nuances

• Experience editing speech or voice recordings (not just music production)

• Strong understanding of speech audio quality standards and common issues in recorded dialogue

• Hands-on experience with tools like iZotope RX, Pro Tools, or similar audio restoration software

• Ability to identify and fix artifacts, inconsistencies, and technical defects in audio

• Experience working with high-volume audio datasets or structured workflows

Preferred Qualifications

• Experience working with TTS vendors or platforms

• Prior involvement in speech ML, voice AI, or dataset creation/annotation workflows

• Familiarity with audio QA pipelines, labeling, or evaluation processes

• Experience directing or editing Brazilian Portuguese voice talent recordings

Ideal Candidate Profile

• Detail-oriented with strong critical listening skills in Brazilian Portuguese

• Comfortable working with repetitive, high-precision tasks at scale

• Able to balance speed and quality in production environments

• Familiar with the nuances of human Brazilian Portuguese speech vs synthetic voice requirements