Wenchen Victor Shi

About Wenchen Victor Shi

Hi everyone! My name is Wenchen (Victor) Shi, and I am a PhD student in Computational Linguistics and Hispanic Linguistics at Indiana University Bloomington, with a background in Applied Mathematics, Spanish, and Linguistics (B.S.+B.A, UCLA) and a Specialization in Computing.I am passionate about phonetics & phonology, speech technology, code-switching, and underrepresented languages, and my recent projects include building automatic speech recognition (ASR) systems for Guarani-Spanish (Jopara) and developing morphological and phonological resources for Indigenous languages in North America.Previously at UCLA, I served as President of the UCLA Spanish Club, Data Analyst Intern at the Data Science Union, Tech Committee Leader for International Bruin’s Business, New Student Advisor for the New Students Academic Program, and Researcher in Spanish Linguistics.Outside of academia, I enjoy making multilingual song covers, exploring sports analytics, and playing Go (Weiqi).

Experience

1. Computational Linguist

   The Language Conservancy

   May 2025 — Present · Bloomington, IN, US

   Interned with the Computational Linguistics and Software & Systems team at TLC on technology to support Indigenous language revitalization, focusing on Crow (Apsáalooke, a Missouri Valley Siouan language) and Cowlitz Coast Salish.Automatic Speech Recognition (ASR) for Crow• Built Python pipelines for “childrenizing” audio (low-pass filtering, warping, speed/formant manipulation) for the Crow Children’s Audio Book.• Performed phonetic analysis using grammar documentation and phoneme inventories; conducted vowel space analysis (F1 vs. F2) and refined IPA alignments using spectrograms, formants, short-time energy, and zero-crossing rates.• Adapted wav2vec2-xlsr-53-espeak-cv-ft for Crow; mapped pseudo-IPA outputs → phonemes → simplified spellings with phonetic features and Levenshtein distance.• Built a real-time Crow Audio Recognition System for bird/plant vocabulary, achieving 80% recognition accuracy and 96% uniqueness.Morphological Tools for Cowlitz• Expanded the Cowlitz Lemmatizer in Apertium with.lexd and.twol rules for morphologically complex forms.• Validated forms in weekly linguist meetings and prepared outputs for HTML display.• Implemented the Cowlitz Phrase Builder: developed preprocessing scripts to convert Apertium outputs into SQL (word class, features, roots, glosses); designed a web interface for users to construct Cowlitz phrases interactively.

Education

• UCLA

  Bachelor of Science - BS, Applied Mathematics with a Specialization in Computing

• Indiana University Bloomington

  Doctor of Philosophy - PhD, Computational Linguistics

• Indiana University Bloomington

  Doctor of Philosophy - PhD, Hispanic Linguistics

• Orangewood Academy

  High School Diploma, High School/Secondary Diplomas and Certificates

• UCLA

  Bachelor of Arts - BA, Spanish and Linguistics