Maria was one of the best simultaneous translators in the business. English to Spanish, real-time, with 98% accuracy. She'd translated for presidents, CEOs, and Nobel laureates. She had never been defeated by a room — until the International Acoustics Symposium. Yes, the irony was noted.

The conference hall had a beautiful PA system that the audience could hear clearly from their seats. But Maria's translation booth received its audio feed from a ceiling mic in the hall, not a direct board feed. The ceiling mic captured the speaker's voice plus every reflection in the room, plus the audience's coughing, plus the HVAC, plus the hum of 400 laptops. The STI of this audio feed was 0.35.

At STI 0.35, approximately 60% of words can be correctly identified by a native speaker listening carefully. Maria was operating at 60% comprehension while simultaneously translating at conversational speed. Her brain was filling gaps, guessing context, and praying. When the speaker made a technical joke, she translated the setup but had to improvise the punchline because she missed two key words. The Spanish-speaking audience laughed at a different joke than the English-speaking audience.

Simultaneous translation requires STI above 0.60 in the source feed. Ceiling mics in reverberant halls cannot achieve this without signal processing.

The Moral: Translation quality starts with source audio quality. Measure the STI of what your translators hear with SonaVyx STI — a direct audio feed or a close-mic eliminates the room from the equation.