Humairah9/Multilingual-Audio-Preprocessing

I processed 50 Hausa audio files for a speech engine project, converting them to .wav, applying noise reduction, and segmenting by pauses. I organized the data and metadata, showcasing my skills in audio preprocessing and multilingual data handling. # Multilingual-Audio-Preprocessing I processed 50 Hausa audio files for a speech engine project, converting them to .wav, applying noise reduction, and segmenting by pauses. I organized the data and metadata, showcasing my skills in audio preprocessing and multilingual data handling. # Table of Content - Introduction - Objectives - Overview - Data Source - Tools Used - Data Cleaning ad Transformation - Methodology - Visual Insights - Visualizations - Conclusion - Recommendations # Introduction With the rebranding of our startup from Otondo Limited to Otondo Team Ltd, and under the direction of the newly appointed Acting Chief Marketing Officer, Oyenibi David Ariyo, our company is reinforcing its commitment to national standards on data privacy and protection. This report presents the end-to-end workflow undertaken during Week 3 of the TDI Challenge: the download, processing, annotation, and analysis of a 250-file audio dataset in line with Nigerian data compliance requirements. # Objectives - Process and prepare a clean, high-quality audio dataset - Segment the audio into short clips for manual transcription - Create structured metadata for each segment - Perform exploratory data analysis - Enable transcription by qualified native speakers - Align all actions with Nigeria's data protection framework # Overview The project’s goal is twofold: - Technical: Prepare and segment audio data for machine learning and transcription purposes. - Regulatory: Ensure compliance with Nigeria's NDPR (Nigeria Data Protection Regulation) through proper handling, documentation, and consent-based transcription efforts. # Data Source - Dataset: A batch of 250 unique audio recordings provided via a secured download link. - Format: Original format in .mp3 - Content: The audios include various native narratives, conversations, and literature readings, primarily in Nigerian indigenous languages. - Confidentiality: Files are anonymized and managed in a privacy-compliant environment. …