Gladia

Gladia delivers speech-to-text and audio intelligence APIs tailored for developers who need accurate, low-latency transcription inside their products. It combines asynchronous batch transcription with real-time streaming, multilingual support, and a growing layer of AI add-ons such as summarization, sentiment analysis, and entity extraction.

What is Gladia? Gladia delivers speech-to-text and audio intelligence APIs tailored for developers who need accurate, low-latency transcription inside their products. It combines asynchronous batch transcription with real-time streaming, multilingual support, and a growing layer of AI add-ons such as summarization, sentiment analysis, and entity extraction. Key Features: Solaria universal STT model: Proprietary Solaria-1 powers transcription across more than 100 languages, handling accents and domain-specific jargon while maintaining high accuracy for names, numbers, and other key entities. Real-time streaming with ultra-low latency: Real-time APIs target latency under 300 milliseconds, with a Partials feature that returns partial transcripts in under 100 milliseconds for live assistants and voice agents. Audio intelligence add-ons: Beyond raw transcripts, Gladia offers diarization, automatic language detection and code-switching, sentiment analysis, named entity recognition, word-level timestamps, summarization, and any-to-any translation. Telephony and communication stack support: The API is tuned for SIP and common telephony protocols at 8 kHz, and works with WebRTC and popular communications platforms, fitting easily into existing voice pipelines.