Building a Production Greek-English Speech Recognizer

• 类型：arxiv
• 标识：2609.13498
• 链接：https://arxiv.org/abs/2609.13498
• 主分类：multimodal
• 形态：position
• TLDR：We report a multi-month engineering program to build Sophea, a production bilingual Greek-English automatic speech recognition system. We evaluate the system against nine production gates covering Greek and English word error rate, language identification, and hallucinations on non-speech audio. Across twenty-three training iterations and two model architectures, no training-data composition passed all nine gates simultaneously. Meeting the Greek noisy-environment target required about 1,500 steps of dense domain exposure, while preserving English language identification tolerated only about 2
• 副分类：engineering
• 待LLM分类：否
• 标题中文：构建生产级希腊语-英语语音识别器
• TLDR中文：我们报告了一项历时数月的工程项目，用于构建 Sophea，一个生产级希腊语-英语双语自动语音识别系统。我们依据九个生产门控评估该系统，涵盖希腊语和英语词错误率、语言识别及非语音音频的幻觉问题。经过二十三次训练迭代和两种模型架构，没有任何训练数据组合能同时通过全部九个门控。满足希腊语噪声环境目标需要约1,500步密集领域暴露，而保持英语语言识别仅能容忍约2%
• 来源文件：
• /inbox/tom/_candidates/2026-09-15-agent-rag-longcontext-candidates.json