Wavix is a communications platform and API provider: phone numbers, SIP trunks, SMS and RCS messaging, call recording, voicemail detection, and speech analytics, sold to businesses and to PBX operators such as 3CX users. For agent builders the relevant piece is Call Media Streaming, which forwards live call audio to your application over a WebSocket so a voice agent can listen and speak on a real phone call.
A stream is started through the API after a call connects, with three parameters: the channel to stream (inbound, outbound, or both), the type (one-way or two-way), and the WebSocket URL. Audio arrives as 20 ms mono PCM16 frames at 24 kHz, base64-encoded in JSON messages. A one-way stream only delivers audio; a two-way stream also accepts audio your agent sends back, so speech-to-text, an LLM, and text-to-speech can be chained on the far end. The feature requires a Flex or Flex Pro account and an API key.
Features
- Call Media Streaming: send live call audio to a WebSocket endpoint in real time
- Bidirectional audio: two-way streams let the agent play audio into the call while receiving the caller's speech
- Message protocol: connected, start, media, stop, and mark messages from Wavix; media, mark, and clear messages from your app
- Audio format: 20 ms mono PCM16 frames at 24 kHz, little-endian, base64-encoded
- Stream limits: one bidirectional plus up to five unidirectional streams per call
- Phone numbers and SIP trunks: local, mobile, and toll-free numbers worldwide, with number porting
- Messaging: SMS, RCS with SMS fallback, 2FA/OTP, and SMS broadcast
- Call tooling: call recording, AI voicemail detection (AMD), and speech analytics
- Account tiers: streaming is available on Flex and Flex Pro accounts with Bearer-token API auth