BeaverTalk: Oregon State University's IWSLT 2025 Simultaneous Speech Translation System

TL;DR

BeaverTalk is a cascaded speech translation system for IWSLT 2025 that combines VAD, Whisper ASR, and Gemma LLMs, fine-tuned with LoRAs for high-quality simultaneous translation in English-German and English-Chinese tasks.

Contribution

This work introduces BeaverTalk, a novel cascaded speech translation system with fine-tuned LLMs and a conversational prompting strategy for improved simultaneous translation performance.

Findings

Achieved BLEU scores of 24.64 and 27.83 for English-German at different latency levels.

Achieved BLEU scores of 34.07 and 37.23 for English-Chinese at different latency levels.

Demonstrated effective integration of VAD, Whisper, and Gemma models for real-time translation.

Abstract

This paper discusses the construction, fine-tuning, and deployment of BeaverTalk, a cascaded system for speech-to-text translation as part of the IWSLT 2025 simultaneous translation task. The system architecture employs a VAD segmenter for breaking a speech stream into segments, Whisper Large V2 for automatic speech recognition (ASR), and Gemma 3 12B for simultaneous translation. Regarding the simultaneous translation LLM, it is fine-tuned via low-rank adaptors (LoRAs) for a conversational prompting strategy that leverages a single prior-sentence memory bank from the source language as context. The cascaded system participated in the English$\rightarrow$German and English$\rightarrow$Chinese language directions for both the low and high latency regimes. In particular, on the English$\rightarrow$German task, the system achieves a BLEU of 24.64 and 27.83 at a StreamLAAL of 1837.86 and 3343.73, respectively. Then, on the English$\rightarrow$Chinese task, the system achieves a BLEU of 34.07 and 37.23 at a StreamLAAL of 2216.99 and 3521.35, respectively.