David AI

David AI is an audio data research company that develops proprietary datasets for speech recognition, translation, synthesis, and conversational AI. Its work is aimed at helping audio models learn capabilities that are difficult to obtain from existing data, using a process that moves from defining a target capability through data design, collection, evaluation, and production-scale release.

The company's featured datasets include Converse, an English dataset of channel-separated two-speaker conversations; Atlas, a multilingual dataset covering more than 15 languages with dialect and accent metadata; Chorus, conversations with three or more speakers designed for speaker separation and diarization; and Dialog, expert conversations across multiple domains. Additional proprietary datasets are available, and the company also works with research teams to design new data shapes for specific use cases.

Access is arranged through a data license agreement covering the required datasets and use cases, with samples provided after an initial call. Off-the-shelf datasets are made available to teams within one to two days of licensing. The company does not publish a free plan, trial, or seat-based pricing structure on the page.