About the Role
The role involves reviewing and validating audio annotations to ensure linguistic accuracy and data quality for artificial intelligence training systems, with a focus on French as spoken in Belgium.
Responsibilities
- Review audio clips recorded in Belgian French
- Verify the correctness of text transcriptions linked to audio samples
- Identify and flag inconsistencies in time-stamped annotations
- Ensure translations and transcriptions adhere to provided guidelines
- Check for proper labeling of speaker attributes and emotions
- Report technical issues with audio files when detected
- Maintain consistency across annotation tasks
- Follow detailed instructions for data categorization
- Validate metadata associated with audio recordings
- Distinguish between dialectal variations and errors
- Assess audio quality and clarity
- Confirm speaker gender and age range when labeled
- Ensure privacy compliance in handling voice data
- Meet quality and throughput benchmarks
- Collaborate with team leads on ambiguous cases
- Provide feedback on annotation guidelines
- Track time spent on tasks accurately
- Adhere to project-specific formatting rules
- Participate in calibration sessions
- Stay updated on changes to project requirements
Nice to Have
- Experience with AI or machine learning data projects
- Knowledge of phonetics or phonology
- Background in dialectology
- Familiarity with speech recognition systems
- Exposure to multilingual data environments
Compensation
Not specified
Work Arrangement
Remote
Team
Distributed team focused on data validation and quality assurance
Project Details
- This is a project-based role focused on validating audio data used to train speech recognition models.
- Work is conducted remotely with flexible scheduling within set deadlines.
Language Requirement
Must be fluent in French as spoken in Belgium, including familiarity with regional expressions and pronunciation.
Technology Setup
- Candidates must have access to a computer with updated browser and high-speed internet.
- No specialized hardware required.
Not applicable
