Build audio and video datasets effortlessly.

Phonic surveys are the easiest way for machine learning researchers to create custom audio and video datasets from scratch.

Proprietary models are built on custom data.

Collect rich responses with audio, video and text prompts in a flexible and dead-simple survey builder.

Scale studies to thousands of responses

Recording audio in either MP3 or multi-channel high fidelity WAV.

Integrate with MTurk and other panel providers

Case Study:
Building a COVID-10 dataset.

During the height of the COVID-19 pandemic, one startup used Phonic to build a model to reliably detect COVID using only the recording of a patient's cough.

Read the case study