Back to Projects

Speech-to-Text Application for Stuttering Individuals

About This Project

Designed a speech to text application that translates stuttered speech into clear, fluent text to support autistic and speech disordered individuals. Analyzed over 28000 audio samples from the SEP 28k dataset and implemented Whisper AI with NumPy and Pandas to train and optimize the model's performance, improving transcription accuracy by 87%.

Key Achievements

  • Analyzed over 28,000 audio samples from the SEP 28k dataset
  • Improved transcription accuracy by 87%
  • Designed to support autistic and speech disordered individuals

Technologies Used

PythonWhisper AIPandasNumPyHugging Face