Welcome to my page

Research Experience

Undergraduate Researcher, Carnegie Mellon University Feb 2022-Jul 2023 Prof Shinji Watanabe’s WAVLab
            Speech Recognition Expertise:
            Independently acquired in-depth knowledge of speech recognition through extensive research.
            Completed Prof. Shinji Watanabe’s course, "Speech Recognition and Understanding," enhancing foundational understanding.
            Utilized the ESPnet speech recognition framework to develop and contribute a
            Bengali speech recognition model to the toolkit on GitHub. 
            Language-Universal Alignments Project:
            Led the creation of a benchmark and a tool for language-universal alignments. Managed 2TB of audio and transcription data from YouTube in 30 languages. Trained speech recognition models for 15/30 languages with varying datasets, showcasing adaptability.
            Presented findings at end-of-semester lab presentations, demonstrating effective
            communication and collaboration skills. JSALT Workshop Project:
            Continued work from the JSALT workshop, training multiple models on synthetic data for the Chinese/English language pair.
            Achieved significant performance improvements compared to training solely on monolingual data.
            Explored various synthetic data types to understand their impact on speech quality and performance.
            Created additional variations for further experimentation.
            Collaborated with another participant to research and author a paper, which was successfully accepted at a peer-reviewed venue.
            JSALT Undergraduate Participant, Johns Hopkins University Jun 2022-Aug 2022 Speech Tech Summer School:
            Completed a rigorous 2-week summer school in speech and language technology,
            staying abreast of cutting-edge advancements. Code-Switched Speech Project Leadership:
            Took the lead on a project targeting improved automatic speech recognition for code- switched speech with limited monolingual training data.
            Focused on the Chinese/English language pair.
            Synthetic Data Generation and Implementation:
            Collaborated and conducted literature review to come up with a method of generation, and implemented an efficient pipeline for generating synthetic audio examples,
            Communication and Presentation:
            Effectively communicated project progress weekly to the workshop team.
            Featured in live-broadcast JSALT 2022 Closing Presentations, showcasing impactful results.
            Data Science Intern, Behaivior Jun 2021 - Aug 2021 Contributions:
            Created custom machine learning models to predict opioid cravings from wearable data
            Created a set of gradient-boosted-tree based classifiers to clean mislabeled data Identified optimal features from timeseries data using a modified Markov Chain Monte Carlo, accounting temporal dependencies
            Increased performance of classifiers by 5-10% using a combination of hyperparameter tuning, data cleaning, and optimal feature selection
            Research Assistant, Carnegie Mellon University May 2020 - Sep 2020 Professor LP Morency’s Lab
            Contributions:
            Using Python, fine-tuned a Factorized-Multimodal-Transformer to classify didactic interactions between parents and children in
            order to eventually detect signs of mental illness
            Used forced-alignment tools to align audio data and their transcripts
            Extracted features from the transcripts in order to feed into the model
            Through preparing for the project, obtained valuable knowledge about state-of-the-art Deep Learning methods and models
            Research Intern, Illinois Institute of Technology Jul 2018 - Sep 2018 Contributions:
            Using Python, manually classified numerous movie reviews into positive and negative for training
            First used a Naive-Bayes Classifier on the unlabeled with a given bank of positive and negative words.
            Later used a CNN model and RNN model to classify the unlabeled data to account for the significance of the order of the words.
            Research Intern, Northwestern University Jun 2018 - Jul 2018 Contributions:
            Visited multiple dermatology practices and photographed images of skin cancers. Worked with a post-doc on an image recognition program to detect skin cancers based on image data.
            Research Intern, George Mason University Jun 2017 - Aug 2017 Contributions:
            Used Autodesk Maya and Matlab to model extraocular muscles in three dimensions. Worked with a PhD student to develop a model for a biomechanical simulator of strabismus.

Dorsa Zeinali

News!

Bio

Education

Honors and Awards

Publications

Conference Papers

Workshop Papers

Service

Research Experience

Skills

Languages