Principal Speech Data Linguist

The role

You will own the linguistic foundation of segmentation and transcription work across languages and domains. You will define standards, establish quality frameworks, and design human-in-the-loop workflows for speech and audio models.

What you will do

• Define transcription and segmentation standards, style guides, and annotation conventions.
• Establish and run quality frameworks including rubrics, error taxonomies, and adjudication processes.
• Own the quality lifecycle for transcription and segmentation deliverables end to end.
• Design human-in-the-loop workflows for human review, correction, and adjudication.
• Handle linguistically hard cases such as accented speech, low-resource audio, and domain jargon.
• Partner with the Speech & Audio Research Scientist to turn model objectives into specifications.
• Train, calibrate, and mentor expert transcribers and reviewers.
• Represent the transcription and segmentation approach to customers and frontier labs.

Required qualifications

• Substantial industry experience in transcription, segmentation, and speech-data quality (typically 8+ years).
• Bachelor's degree in linguistics, phonetics, or computational linguistics, or a closely related field.
• Strong foundation in phonetics, phonology, and sociolinguistics.
• Fluency in phonetic transcription and IPA.
• Hands-on experience with acoustic and phonetic analysis of speech using tools like Praat.
• Deep experience with audio segmentation and its conventions.
• Hands-on fluency with the modern speech stack including Whisper, commercial ASR engines, forced alignment, and annotation tools like ELAN.
• Comfort scripting for speech-data work using Python and regular expressions.
• Practical data-management skills across the delivery lifecycle.
• Multilingual capability and hands-on experience with accented, dialectal, and code-switched speech.
• Strong written and verbal communication.

Nice to have

• An advanced degree
• Low-resource languages
• Responsible-AI considerations for speech, such as bias across accents and dialects and privacy and consent in voice data

About the company

Innodata is a global data engineering company that provides data, evaluation frameworks, and human expertise to build AI systems. They have a 36+ year legacy delivering data and outcomes for customers.