Real-time Automatic Speech Recognition (ASR) is a fundamental building block for many commercial applications of ML, including live captioning, dictation, meeting transcriptions, and medical scribes. Accuracy and latency are the most important factors when companies select a system to deploy. We present WhisperKit, an optimized on-device inference system for real-time ASR that significantly outperforms leading cloud-based systems. We benchmark against server-side systems that deploy a diverse set of models, including a frontier model (OpenAI gpt-4o-transcribe), a proprietary model (Deepgram nova-3), and an open-source model (Fireworks large-v3-turbo).Our results show that WhisperKit matches the lowest latency at 0.46s while achieving the highest accuracy 2.2% WER. The optimizations behind the WhisperKit system are described in detail in this paper.
@article{arxiv.2507.10860,
title = {WhisperKit: On-device Real-time ASR with Billion-Scale Transformers},
author = {Atila Orhon and Arda Okan and Berkin Durmus and Zach Nagengast and Eduardo Pacheco},
journal= {arXiv preprint arXiv:2507.10860},
year = {2025}
}
Comments
ICML 2025 - On-Device Learning for Foundational Models Workshop