Voice AI Agents Struggle With Bilingual Speech, Study Finds
A new study from ServiceNow-AI found that most voice agents struggle to understand people who speak two languages at once (code-switching). The team tested 7 speech recognition systems on Spanish-English, French-English and other language pairs. ElevenLabs Scribe V2 and AssemblyAI Universal-3 Pro delivered the best results, while OpenAI Whisper simply translates bilingual speech instead of transcribing it.
AI-processed from Hugging Face Blog; edited by Hamidun News
Voice AI agents handle bilingual customers poorly, according to a study by ServiceNow-AI. The company tested seven popular speech recognition systems on examples of code-switching, where people move smoothly from one language to another within a single utterance.
What is code-switching in speech?
Code-switching is when people move smoothly from one language to another within a single utterance, weaving words and phrases together.
Why do voice AI agents handle bilingualism poorly?
Voice assistants and agents are trained mainly on monolingual data, so they struggle to recognize mixed speech.
Key Questions Answered
What is code-switching in speech?
Code-switching is when people move smoothly from one language to another within a single utterance, weaving words and phrases together.
Why do voice AI agents handle bilingualism poorly?
Voice assistants and agents are trained mainly on monolingual data, so they struggle to recognize mixed speech.
Test Frontier LLMs & Autonomous Agents in One Workspace
Run Claude Opus 5.5, GPT-6, Gemini 3.1, and DeepSeek side-by-side in Nomad AI (nomadnet.ai) — unified model routing, deep research, and multimodal workflows under one subscription.
The AI world, distilled — once a week
Seven stories that actually mattered, hand-picked. No noise, no reposts, no press releases.
Done! Check your inbox for a confirmation.