Sanas focused on contact center accent conversion

The key advantage is not that Sanas does more voice AI, it is that Sanas solves one ugly telecom problem the horizontal platforms are not built around. In a live support call, the product has to sit in the audio path, preserve the agent’s identity, work on 8kHz phone audio, and keep delay low enough that turn taking still feels natural. Deepgram, OpenAI, and ElevenLabs all cover adjacent layers, but their products are broader speech and agent stacks rather than a dedicated accent conversion workflow for contact center calls.1 2 3 4 5 6

• Sanas is tuned for the contact center deployment model. Its help docs show Accent Translation built around softphone and headset configuration, and its AT 4.6.0 model added 8kHz support specifically for low bandwidth telephony. That is very different from a generic developer API that assumes an app builder will stitch the call stack together.1 5 6
• Deepgram has gone horizontal in the other direction. Its Voice Agent API bundles speech to text, language model orchestration, and text to speech over one WebSocket, with Twilio guides for phone agents and optional bring your own TTS. That makes it strong infrastructure for building call automation, but not a packaged accent conversion product for human agents already on the line.2 7 8
• ElevenLabs and OpenAI are also broad platforms. ElevenLabs offers phone integrations for conversational agents and custom telephony audio settings, while OpenAI Realtime supports speech to speech sessions over WebRTC, WebSocket, and SIP. Those products are designed to power AI conversations and voice apps at scale, not to quietly rewrite a human agent’s accent inside an existing contact center workflow.3 4 9 10 11

Over time, the market is likely to split between broad voice platforms and narrow workflow products. Sanas is positioned to win where buyers want a drop in layer for offshore support teams, then expand into noise reduction, translation, and intelligence. The next competitive step is horizontal vendors moving down into telephony grade accent conversion once contact center demand is proven.1 2 3 4