# Contrastive Feedback Mechanism for Simultaneous Speech Translation Source: [Contrastive Feedback Mechanism for Simultaneous Speech Translation]() Collection index: [llms.txt]() Data updated: 2026-04-14T05:00:04.234Z Authors: Haotian Tan, Sakriani Sakti Published: 2024-07-30T03:50:10.000Z ## Abstract Recent advances in simultaneous speech translation (SST) focus on the decision policies that enable the use of offline-trained ST models for simultaneous inference. These decision policies not only control the quality-latency trade-off in SST but also mitigate the impact of unstable predictions on translation quality by delaying translation for more context or discarding these predictions through stable hypothesis detection. However, these policies often overlook the potential benefits of utilizing unstable predictions. We introduce the contrastive feedback mechanism (CFM) for SST, a novel method that leverages these unstable predictions as feedback to improve translation quality. CFM guides the system to eliminate undesired model behaviors from these predictions through a contrastive objective. The experiments on 3 state-of-the-art decision policies across 8 languages in the MuST-C v1.0 dataset show that CFM effectively improves the performance of SST. ## Recorded benchmark results A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage. - [Simultaneous Speech Translation on MuST-C en-nl 1.0 (test)](): Simultaneous Speech Translation; MuST-C (En-Nl) 1.0 (test); BLEU: 31.41. - [Simultaneous Speech Translation on MuST-C en-pt v1.0 (test)](): Simultaneous Speech Translation; MuST-C en-pt v1.0 (test); BLEU: 32.23. - [Simultaneous Speech Translation on MuST-C en-ro 1.0 (test)](): Simultaneous Speech Translation; MuST-C En-Ro 1.0 (test); BLEU: 26.44. - [Simultaneous Speech Translation on MuST-C en-ru v1.0 (test)](): Simultaneous Speech Translation; MuST-C en-ru v1.0 (test); BLEU: 18.5. - [Simultaneous Speech Translation on MuST-C en-de v1.0 (test)](): Simultaneous Speech Translation; MuST-C en-de v1.0 (test); BLEU: 27.38. - [Simultaneous Speech Translation on MuST-C en-es v1.0 (test)](): Simultaneous Speech Translation; MuST-C En-Es v1.0 (test); BLEU: 27.18. - [Simultaneous Speech Translation on MuST-C en-fr v1.0 (test)](): Simultaneous Speech Translation; MuST-C en-fr v1.0 (test); BLEU Score: 33.92. - [Simultaneous Speech Translation on MuST-C en-it v1.0 (test)](): Simultaneous Speech Translation; MuST-C en-it v1.0 (test); BLEU: 23.14. ## Original sources and implementation [Original paper]() [PDF]()