Paper Overview
Field: NLP Authors: Vladimir Bataev, Lilit Grigoryan, Andrei Andrusenko, Nikolay Karpov, Vitaly Lavrukhin, Boris Ginsburg Published: 2026-08-21 arXiv: 2608.21343
Abstract
Contextualization is essential for production automatic speech recognition (ASR) systems, where user-provided phrases must be recognized accurately under strict latency constraints. Although many context-biasing methods improve recognition accuracy, they often do not address the practical requirements of modern production ASR systems: streaming inference, efficient batched decoding, user-specific context lists, and low runtime overhead.
The authors propose TurboBias 2.0, a production-oriented framework for efficient phrase boosting in Transducer-based ASR systems. The framework extends GPU-accelerated TurboBias with a case-insensitive boosting graph and per-stream batched decoding, allowing each utterance in a batch to use an independent context-biasing configuration. This enables personalized context biasing for multiple simultaneous users without sharing or mixing their context lists.
The proposed framework supports both offline and streaming inference and can be used with greedy search and beam search decoding. Experiments demonstrate that TurboBias 2.0 improves contextual phrase recognition while maintaining low latency and high throughput.
Key Points
- Extends GPU-accelerated TurboBias with a case-insensitive boosting graph
- Per-stream batched decoding enables independent context configurations per utterance in a batch
- Supports personalized context biasing for multiple concurrent users without mixing context lists
- Works with both offline and streaming inference, and with greedy and beam search decoding
- Improves contextual phrase recognition while keeping low latency and high throughput
*Auto-collected on 2026-08-25. Source: arXiv:2608.21343*