Datasets:
license: cc-by-4.0
language:
- en
tags:
- audio
- speech
Data Description
This dataset contains 50 conversational English sentences recorded on my phone in a quiet environment. The sentences cover everyday topics such as daily routines, productivity, weather, relaxation habits, and short reflections. All recordings were segmented into individual WAV files named sequentially from s001.wav to s050.wav. The metadata.csv file links each audio clip to it's corresponding transcription using the pipe-delimited format.
Issues Encountered and Solutions
During preparation, I encountered two main issues. The first was ensuring that every audio file name matched correctly with the entries in my metadata.csv, especially after reorganizing the files and cleaning the naming format. The second issue involved dealing with formatting and structural requirements for the Hugging Face Dataset Viewer, including adjusting metadata placement and correcting delimiters so the dataset could be parsed properly. To resolve these problems, I stepped away periodically and reviewed the dataset with a fresh mind, making careful corrections until everything aligned cleanly and the repository structure functioned as intended.