ASR Linguistics Intern (AI)

TL;DR

ASR Linguistics Intern (AI): Researching and optimizing text normalization and formatting for ASR systems in automotive voice assistants with an accent on tokenizer and formatter accuracy. Focus on evaluating LLMs versus rule-based grammars for multilingual text processing and the handling of complex linguistic phenomena.

Location: Must be based in Aachen, Germany

Company

Global industry leader in creating unique voice and AI experiences for the automotive world, working with leading automakers like BMW, Audi, and Ford.

What you will do

• Research text normalization (tokenization) and inverse normalization (formatting) for statistical and neural network ASR systems.
• Evaluate whether LLMs can achieve formatting accuracy comparable to rule-based grammars.
• Analyze the performance of multilingual systems and determine the amount of training data required for high accuracy.
• Develop methods to handle complex phenomena including numerals, time expressions, dates, and street addresses.
• Experiment with dividing the system into concept identification and actual formatting stages.

Requirements

• Background or strong interest in linguistics and ASR systems.
• Ability to work with monolingual and multilingual systems.
• Analytical mindset to conduct research on model accuracy and training data.
• Location: Based in Aachen, Germany

Culture & Benefits

• Opportunity to contribute to meaningful innovation in a rapidly growing AI industry.
• Work within a passionate, dedicated global team.
• Experience building the future of voice assistants for the world's leading automotive brands.