--- title: "Google and ElevenLabs voice AI miss up to 38% of new drug names in Synthio Labs' public benchmark" description: "A Y Combinator-backed startup's public benchmark found general-purpose voice models miss up to one in three newly approved drug names, and pharma is deploying those agents now." publisher: "The Inference" section: "Model" published: 2026-11-07T16:17:16.468Z modified: 2026-11-07T16:17:16.468Z canonical: https://theinference.org/article/google-and-elevenlabs-voice-ai-miss-up-to-38-of-new-drug-names-in-synthio-labs-public-benchmark language: en keywords: "Voice AI, Benchmarks, Healthcare AI, Pharma" --- # Google and ElevenLabs voice AI miss up to 38% of new drug names in Synthio Labs' public benchmark > A Y Combinator-backed startup's public benchmark found general-purpose voice models miss up to one in three newly approved drug names, and pharma is deploying those agents now. Voice agents are moving into pharmaceutical engagement this month: on November 5, 2026, [Synthio Labs](https://theinference.org/markets/companies/synthio-labs), a San Francisco startup selling clinical-grade voice AI to pharma, published [five takeaways from a Fierce Pharma Week fireside chat](https://synthiolabs.com/resources) it shared with Genentech [4]. Two months earlier, on September 17, 2026, the same company [published the accuracy test those agents will be judged against](https://finance.yahoo.com/healthcare/articles/synthio-labs-launches-dose-benchmark-120000124.html), and the general models fail it [1]. The test is DOSE (Drug-name Oral Synthesis Evaluation), the first public benchmark of how well text-to-speech systems pronounce drug names. ## The test, and who failed it DOSE scored nine commercial text-to-speech models, including systems from [ElevenLabs](https://theinference.org/markets/companies/elevenlabs), Google, Deepgram, [Microsoft](https://theinference.org/markets/companies/microsoft) and [OpenAI](https://theinference.org/markets/companies/openai), on 274 drug names, 146 of them newly approved [1]. Each name was read inside a real clinical sentence and graded 0 to 5 against verified pronunciations; a score of 4 or higher passes [1]. General-purpose models passed between 63.1% and 80.3% of names overall [1]. On new drugs every model dropped: ElevenLabs' eleven_v3 fell from 93.0% on established names to 67.1%, [Google](https://theinference.org/markets/companies/alphabet)'s Gemini TTS from 89.1% to 61.6% [1]. Microsoft Azure passed fewer than half of generic names (INNs, International Nonproprietary Names, the standard generic drug names) [1]. One system spelled the flu drug Xofluza letter by letter [1]. *Chart: **New drug names are where general voice AI breaks** Column chart of pronunciation pass rates on established versus newly approved drug names for ElevenLabs eleven_v3 (93.0 vs 67.1) and Google Gemini TTS (89.1 vs 61.6), from the DOSE benchmark. Synthio's RxPronounce has no published established-name score; on newly approved names it passed 87.0%, 10.9 points ahead of the next system.* *Pass rates on DOSE, the drug-name benchmark Synthio Labs published on September 17, 2026, percentage of 274 drug names graded 4 or higher on a 0-to-5 scale; no established-name figure is published for Synthio's RxPronounce, whose newly approved pass rate leads the next system by 10.9 points [1]. [1]* Synthio's own RxPronounce model topped the table at 91.2% overall and 87.0% on newly approved names, 10.9 points ahead of the next system [1]. ## An old safety category, a new speaker Rajashekar Vasantha, Synthio's co-founder and CTO, whose team built DOSE, argues the safety category is old even if the benchmark is new: "Sound-alike drug name confusion is a patient-safety category WHO, ISMP and FDA already track" [1]. [Independent record confirms the Institute for Safe Medication Practices](https://en.wikipedia.org/wiki/Institute_for_Safe_Medication_Practices) (ISMP), a nonprofit founded in 1975, has spent decades investigating and publishing medication errors that occurred in practice [5]. Voice AI adds a new speaker to that problem. Chief executive Supreet Deshpande supplied the mechanism: "The pattern tracks where training data runs out" [1]. Every model handles metformin, a legacy diabetes drug, but struggles with the molecule approved last quarter, exactly the name a launch team or patient most needs spoken correctly [1]. ## A vendor's benchmark crowns its own model The caveat is structural: this is a vendor's benchmark crowning the vendor's own model. [Synthio's research index dates DOSE to August 2026](https://synthiolabs.com/research), while the announcement went out on September 17 [2][1]. The full dataset, audio and scoring rubric are published, on [Hugging Face](https://theinference.org/markets/companies/hugging-face) among others [1][2], but no independent re-run exists yet. Watch for one. If the 10.9-point gap survives outside review, vertical pronunciation layers become the wedge that displaces general voice APIs in [regulated healthcare](https://theinference.org/article/hippocratic-ai-prices-healthcare-voice-ai-by-risk-tier-from-2-50-an-hour-for-admin-calls-to-15), and Synthio, already live at several top-10 pharma companies and on [AWS](https://theinference.org/markets/companies/amazon) Marketplace, fields the product to collect [1][4]. ## Takeaway General voice models pass only 61.6% to 67.1% of newly approved drug names on the DOSE benchmark, and if Synthio's 10.9-point lead survives independent review, vertical pronunciation layers become the wedge against general voice APIs in regulated healthcare. ## Sources 1. [CNW Group via Yahoo Finance, Synthio Labs Launches DOSE Benchmark: Leading Voice AI Models Mispronounce Up to One in Three Newly Approved Drug Names, 17 September 2026](https://finance.yahoo.com/healthcare/articles/synthio-labs-launches-dose-benchmark-120000124.html) 2. [Synthio Labs, Research: benchmarks, studies and engineering (DOSE Benchmark v1, dataset and rubric index)](https://synthiolabs.com/research) 4. [Synthio Labs, Resources index: Reaching HCPs with AI, 5 takeaways from our Fierce Pharma Week fireside chat with Genentech, 5 November 2026](https://synthiolabs.com/resources) 5. [Wikipedia, Institute for Safe Medication Practices, accessed 7 November 2026](https://en.wikipedia.org/wiki/Institute_for_Safe_Medication_Practices) --- Google and ElevenLabs voice AI miss up to 38% of new drug names in Synthio Labs' public benchmark — The Inference. Canonical: https://theinference.org/article/google-and-elevenlabs-voice-ai-miss-up-to-38-of-new-drug-names-in-synthio-labs-public-benchmark