Location: Remote (Vietnam Preferred)
Employment Type: Contract / Full-time
Work Hours: 40 hours per week (8 hours per day)
Start Date: Immediate / As soon as possible
Job Overview
We are seeking a Vietnamese ASR Linguist (Automatic Speech Recognition Linguist) to support the development, training, and evaluation of speech recognition and natural language understanding systems.
This role combines Vietnamese linguistic expertise with applied NLP and ASR technologies. The selected candidate will work closely with AI researchers, data scientists, and speech engineers to improve language models, pronunciation lexicons, acoustic models, and conversational AI systems.
The role involves linguistic analysis, dataset preparation, annotation guideline creation, model evaluation, and error analysis to improve the performance and accuracy of Vietnamese language technologies.
Key Responsibilities
- Analyze Vietnamese language data to support Natural Language Understanding (NLU) and speech recognition model development
- Design, maintain, and validate pronunciation lexicons and phoneme inventories
- Perform sentence structure analysis, linguistic evaluation, and intent classification
- Develop and maintain annotation guidelines and transcription standards
- Perform quality control of linguistic datasets, including transcription validation and lexicon verification
- Conduct error analysis on model outputs and identify systematic language model or recognition issues
- Support data preparation pipelines for ASR model training and evaluation
- Generate phoneme-level alignments and linguistic annotations for training speech models
- Work closely with engineers and data scientists to improve language model accuracy and speech recognition performance
- Document linguistic patterns, edge cases, and recommended improvements for AI models
Required Qualifications: Language Skills
- Strong proficiency in Vietnamese
- Deep understanding of Vietnamese grammar, phonology, and linguistic structures
- Strong written and verbal communication skills in Vietnamese
Education
Bachelor's or master's degree in one of the following fields:
- Computer Science
- Computer Engineering
- Computational Linguistics
- Artificial Intelligence
- Linguistics or a related technical discipline
Skills Required: Core Linguistic Skills (Vietnamese)
- Strong knowledge of Vietnamese phonology, including phoneme inventory, tones, syllable structure, and regional variants (e.g., northern vs southern pronunciations)
- Understanding of common pronunciation confusions and phonetic variations
- Ability to design and maintain pronunciation lexicons and phoneme sets
- Understanding of Vietnamese morphology and word segmentation to support text normalization and language model training
- compound structures
- clitics
- reduplication patterns
- Experience defining annotation guidelines and transcription standards
- Ability to perform or lead speech transcription and linguistic annotation workflows
- Strong quality control skills for transcription and lexicon validation
- spot-checking annotations
- inter-annotator agreement analysis
- error pattern analysis
Acoustic Model / ASR Technical Skills
- Solid understanding of statistical acoustic modeling concepts, including:
- HMM/GMM models
- hybrid HMM-DNN architectures
- context-dependent phones (triphones/senones)
- decision trees
- Knowledge of techniques for extracting speech features, such as:
- MFCC
- Filterbank (Fbank) features
- Familiarity with speaker adaptation techniques
- Experience with ASR (Automatic Speech Recognition) toolkits such as Kaldi for data preparation and alignment.
- data preparation
- alignment
- acoustic model training
- decoding
- Experience performing forced alignment and label generation
- phone-level alignments
- pronunciation variants
- short pauses
- Ability to design ASR (Automatic Speech Recognition) training recipes, including:
- dataset splits
- data augmentation strategies
- Experience with model evaluation and debugging, including:
- Word Error Rate (WER)
- Character Error Rate (CER)
- Phone Error Rate (PER)
- confusion matrix analysis
Technical Skills
- Working knowledge of Python
- Familiarity with data processing pipelines and dataset preparation
- Experience performing AI/NLP model evaluation and error analysis
- Understanding of machine learning workflows for speech and language models
Equal Opportunity
The employer is an equal opportunity employer and considers all qualified applicants without regard to race, color, religion, sex, sexual orientation, gender identity, national origin, disability status, protected veteran status, or other characteristics protected by law.
