Soniox

Soniox is a single API that handles speech-to-text, text-to-speech and real-time translation across more than 60 languages and some 3,600 language pairs. It’s aimed at use cases where audio can be messy or mixed — think multiple speakers, background noise, foreign names and even mid-sentence switches between languages.

The service focuses on low latency, with streaming transcription under 200ms and the ability to begin generating speech from the first few words. The same models and API can be deployed globally, with in-region processing options to address latency, data residency and regulatory requirements.

• Multilingual coverage: 60+ languages and ~3,600 language pairs
• Handles mid-sentence language switching, alphanumerics, foreign names and multi-speaker audio
• Low-latency streaming, sub-200ms transcription and early speech generation
• Global deployment with in-region processing for data residency and regulatory needs

Organisations such as Perplexity, Samsung, LG, Wonderful and DeliverHealth are listed as users. More info is available at https://soniox.com/.