Real-time and async English speech-to-text API that keeps up with how people actually talk. Handles regional accents, fast speech, and code-switching mid-sentence. Live in production in under a day.
We evaluated Gladia's Word Error Rate (WER) for English across various datasets.
Gladia's internal production dataset of real customer calls, annotated by humans.
Switchboard contains spontaneous telephone conversations between strangers, with interruptions, informal language, and multiple speakers.
Earnings22 consists of corporate earnings calls and financial presentations.
1.4B+
English is the most spoken language in the world once second-language speakers are counted, and most of its speakers learned it as a second language. It's the default language of international business, from the US and UK to India, Nigeria, and the Philippines.
Supporting English lets you serve customers in the US, the UK, Canada, and Australia, plus the far larger group of second-language speakers in India, Nigeria, and the Philippines, with every accent in between.
People rarely stick to one language. Gladia follows English speakers when they switch mid-sentence, so your transcripts stay clean.
Code-switchingUse cases
Whatever you're building, English shouldn't be the thing that slows it down.
International teams use English as their shared working language, so meetings mix accents from Bangalore to Berlin to Lagos in a single call.
Meeting assistants →Contact centers in the Philippines and India handle English support for US and UK brands, with Philippine contact center revenue topping $33B in 2025.
CCaaS →English callers range from Texan to Scottish to Nigerian accents, so voice agents can't assume one standard way of saying anything.
Voice agents →English is the leading language of web content, so podcasts, courses, and videos need transcripts that search engines and global viewers can use.
Media →Supported languages
Being multilingual has been our priority since day one. Gladia supports 100 languages, with code-switching and automatic language detection integrated into all models by default.
Ship English voice features with an API built for messy, real-world conversations.
