Reverb: Open-Source ASR and Diarization from Rev

TL;DR

Reverb provides open-source speech recognition and diarization models, including a full pipeline and research versions, aiming to accelerate innovation and outperform existing open-source models in long-form speech tasks.

Contribution

It introduces open-source, high-performance speech recognition and diarization models with both production and research versions for the first time.

Findings

Models outperform all existing open-source speech recognition models.

Reverb enables research and development in voice technology.

Open-source release encourages community-driven innovation.

Abstract

Today, we are open-sourcing our core speech recognition and diarization models for non-commercial use. We are releasing both a full production pipeline for developers as well as pared-down research models for experimentation. Rev hopes that these releases will spur research and innovation in the fast-moving domain of voice technology. The speech recognition models released today outperform all existing open source speech recognition models across a variety of long-form speech recognition domains.