Why Recording Toilet Flush Sounds?
One of our most unexpected yet popular projects to date involved collecting toilet flushing sounds for a medical application. The client, a health-tech company, was developing an app designed to help clinicians identify signs of diabetes based on patients’ urination frequency.
While increased urination is a known symptom of diabetes, our role wasn’t to validate medical claims. Instead, our objective was to build a high-quality dataset that could help the app accurately detect the sound of a toilet flush, even in varying environments.
The Challenge | Is That a Flush?
To train the app’s AI to recognize flushing sounds reliably, we needed to capture a wide range of acoustic scenarios:
- Bathroom door: open, closed, and partially ajar
- Toilet lid: open and closed
- Varied water pressures and toilet types
- Bathrooms of different sizes and acoustics
- Different door thicknesses for closed-door recordings
This wide variability required careful planning to ensure all relevant scenarios were represented.
The Process | May We Use Your Toilet?
With a clearly defined set of parameters, we began recruiting participants. The response was overwhelming; people were genuinely enthusiastic about contributing to such a unique project. We successfully gathered a robust dataset of toilet flushing sounds. Each recording was carefully reviewed, categorized, and annotated by our expert team before being used to train the AI.
Post-Processing | Flushing With Data
To ensure data quality:
- Poorly recorded or faint audio samples were excluded
- Flush sounds were isolated
- Ambient sounds not related to the flush were removed where possible
- Unusable recordings were discarded
The end result was a clean, rich dataset that offered comprehensive acoustic coverage for the client’s needs.
Outcome | Flushed Success
This quirky yet technically demanding project turned into one of our biggest success stories. The client received a dataset that exceeded expectations, enabling them to build a powerful and reliable AI component for their app— supporting diabetes research and patient care. Internally, the project enriched our audio collection expertise and expanded our capabilities, adding another dimension to our service offerings.
Our Data Services
We specialize in collecting and preparing a wide range of data types essential for AI development, including text data for Natural Language Processing (NLP), speech and audio, image, video, and multimodal data that combines multiple formats.
Learn more about our services on our Data Services page.
Or contact us below to see how you can be our next success story.
