Google Research· ResearchResearch paper

​​Speech-to-Retrieval (S2R): A new approach to voice search

Google Research introduces Speech-to-Retrieval (S2R), a new approach to voice search that directly interprets spoken queries without transcribing them to text first, and open-sources the Simple Voice Questions (SVQ) dataset for evaluating S2R performance.

From the source

In addition, we are open-sourcing the Simple Voice Questions (SVQ) dataset, a collection of short audio questions recorded in 17 different languages and 26 locales, which we used to evaluate the performance potential of S2R.

research.google

Announced

07 Oct 2025 15:22 UTC

Detected

03 Aug 2026 21:12 UTC (+300d after)