PubSpeaker
VISIT WEBSITE
Education
is a speech analysis and feedback application designed to help users improve their public speaking and pronunciation skills. The app transcribes spoken input into text, analyzes pronunciation, grammar, and fluency, and highlights errors or areas for improvement. Users can review past speech results through a history feature, allowing them to reread transcripts or replay their recorded speech for self-evaluation.
We delivered a comprehensive speech analysis experience: accurate speech-to-text transcription, phoneme-level pronunciation evaluation, grammar and fluency analysis, clear visual error highlighting, audio playback for self-review, and a structured history system for tracking past results. This is a complete skill-development solution, designed to help users refine their public speaking abilities with practical, data-driven feedback and continuous improvement support.
integrates speech-to-text transcription, phoneme-level pronunciation analysis, and grammar evaluation to deliver clear, actionable feedback that helps users improve their public speaking skills.
Integrated OpenAI Whisper for speech-to-text transcription and Montreal Forced Aligner (MFA) for phoneme-level alignment to accurately analyze pronunciation.
Applied Word Error Rate (WER) and text alignment techniques to evaluate transcription accuracy and identify pronunciation and fluency issues..
Designed a modular system that transforms transcription and alignment results into clear visual feedback, with audio replay and history tracking for continuous improvement.
Have a project in mind? I'd love to hear about it and discuss how we can collaborate to create something amazing.