Speech Data Labelling Specialist
Job Description
The Speech Data Labelling Specialist is responsible for preparing, classifying, and annotating speech datasets used to develop and improve automatic speech recognition systems, voice-controlled applications, conversational AI platforms, and speech-to-text technologies. The role focuses specifically on identifying spoken language elements, segmenting speech recordings, assigning speaker labels, and ensuring that spoken content is accurately represented in structured datasets.
The specialist will work with speech recordings from different speakers, accents, languages, and communication environments. Responsibilities include identifying pronunciation variations, speech interruptions, disfluencies, pauses, and other linguistic characteristics that influence the performance of speech recognition systems. The position requires strong linguistic awareness, exceptional listening abilities, and a high degree of consistency.
Key Responsibilities:
Listen to and label speech recordings from different speakers.
Produce accurate speech transcripts based on project guidelines.
Identify individual speakers and assign speaker identification labels.
Segment recordings into words, phrases, and sentence-level timestamps.
Identify accents, dialect variations, and pronunciation differences.
Annotate speech interruptions, pauses, repetitions, and filler words.
Identify unclear speech and mark unintelligible audio sections appropriately.
Review speech labels to ensure consistency and transcription accuracy.
Support the development of multilingual speech recognition datasets.
Work with AI engineers to improve speech recognition data quality.
Qualifications and Requirements:
Bachelor’s degree in Linguistics, Language Studies, Computer Science, or a related discipline.
Experience in transcription, speech annotation, or linguistic data processing.
Excellent listening, spelling, and grammar skills.
Understanding of phonetics, pronunciation, and speech segmentation.
Familiarity with speech annotation and transcription software.
Strong attention to detail and ability to handle large audio datasets.
