Skip to content
ITENFRES
Create a free account

Voice API · WebSocket Media Streaming

The audio bridge
between the call
and your AI.

AudioBridge opens a WebSocket on the call and delivers the audio to you in real time, both ways. Very similar structure to Twilio Media Streams — migrating is extremely simple and fast — on the network of an Italian operator: native IT and FR numbers, data stays in the EU, carrier pricing.

Start for free Go to the documentation

100 free minutes on sign-up · no credit card · Node & Python SDKs

Full-duplex, in real time. The switch sends you the caller's voice (uplink) and plays back the audio you send it (downlink). Binary audio or base64 JSON, μ-law or A-law, 20 ms frames.

  • connected
  • start
  • media
  • dtmf
  • mark
  • clear
  • stop

Why AudioBridge

Three things the others cannot say.

01

Run by an operator, not a reseller.

BEEVOIP is a ROC-licensed VoIP operator with 232 Italian area codes. The Italian numbers are ours; the French ones we sign in the MAN. No resale markup: you pay the minute at the carrier's price.

02

Your data stays in Europe.

EU-native infrastructure on Azure Italy and France: call audio never leaves the EU. GDPR DPA signed within 24 hours, public sub-processor list, Art. 30 register ready. Aligned with the AI Act for conversational systems.

03

You choose the AI stack.

AudioBridge is the audio stream: on top you plug in OpenAI Realtime, Gemini Live, Whisper, Deepgram, ElevenLabs, Azure Speech. Ready-made examples for each on GitHub, no model lock-in.

The protocol

One panel to
switch on the bridge.

You configure the project from the audiobridge.beevoip.it portal: point the WebSocket wherever you want, pick the transport and the DTMF. No files to write. The event structure is very similar to Twilio Media Streams, with two additions: binary audio and the A‑law codec.

Documentation →
audiobridge.beevoip.it
{
  "configType": "audio-bridge",
  "wsURL": "wss://stream.tuodominio.it/audio",
  "transport": "binary",
  "dtmfOptions": { "minLength": 1, "timeout": 4000 },
  "wsFrameDuration": 20
}
connected→start→media ⇄→stop

Migration

On Twilio?
Keep your server.

The protocol has a very similar structure to Twilio Media Streams, with a few targeted simplifications: migrating is extremely simple and fast. Your WebSocket server stays the same — same URL, same code. Only the carrier delivering the audio changes.

wss://il‑tuo‑server.it/audio = wss://il‑tuo‑server.it/audio

Carrier-grade

ROC operatorBEEVOIP S.r.l., ministerial licence
232 Italian area codesnative IT geographic numbers
All French area codesnative FR geographic numbers
MAN / STIR‑SHAKENauthenticated French CLI — where US CPaaS cannot reach
Azure IT + FRcall audio always stays in the EU

What you build with it

One audio stream, a thousand applications.

  • Conversational AI voicebotnative barge‑in with the clear event
  • Transfer to an agentto an extension on your PBX (BYOC with SIP trunk) or to a fixed/mobile number
  • Live transcriptionWhisper, Deepgram, Azure Speech
  • Sentiment analysisin real time, while the customer speaks
  • Recording on your sideyou receive the full audio stream: you record it, we keep no audio
  • Conversational IVRvoice menus that understand natural language

Transparent pricing

Carrier pricing, in euro.

Inbound & real time

per-minute billing, prepaid
Inbound on a Beevoip number€ 0.0050/min
WebSocket streaming€ 0.0045/min
BYOC inbound — bring your own numbers€ 0.0050/min
Number fee€ 3.00/mo

Outbound

per-minute billing, prepaid
Italy fixed€ 0.0175/min
Italy mobile€ 0.0450/min
France fixed€ 0.018/min
France mobile€ 0.048/min

Same rates for call transfer to mobile numbers.

Outbound BYOC

per-minute billing, prepaid
BYOC outbound — use your own numbers€ 0.0050/min
WebSocket streaming€ 0.0045/min
Concurrency1 channel = 1 simultaneous call
1channel€ 18.50/mo
2channels€ 36/mo
5channels€ 90/mo
10channels€ 169/mo
10+channelson request

Prices negotiable on volume.

Frequently asked questions

Frequently asked questions about AudioBridge.

What is AudioBridge?

AudioBridge is a Voice API that opens a WebSocket on the call and delivers the audio in real time, both ways (full-duplex). The structure is very similar to Twilio Media Streams and migrating is extremely simple and fast; it runs on the network of BEEVOIP, a ROC-licensed Italian VoIP operator, with native Italian and French numbers.

Is it easy to migrate from Twilio Media Streams?

Yes, it is extremely simple and fast: the event structure is very similar to Twilio Media Streams, with two additions (binary audio and the A-law codec). Migrating from Twilio you keep the same WebSocket server, the same URL and the same code; only the carrier delivering the audio changes.

Where is call data processed?

The infrastructure runs on Azure Italy and France and call audio never leaves the European Union. BEEVOIP provides a GDPR DPA signed within 24 hours, a public sub-processor list and an Art. 30 register.

Which AI stacks can I connect?

AudioBridge is the audio stream: on top you can connect OpenAI Realtime, Gemini Live, Whisper, Deepgram, ElevenLabs or Azure Speech. GitHub has ready-made examples for each, with no model lock-in.

What audio format does AudioBridge use?

Audio at 8000 Hz, mono. The codec is A-law by default on BEEVOIP numbers, with µ-law negotiable on trunks. The audio travels as raw codec bytes (transport binary) or base64 (transport json), in 20 ms frames.

How much does AudioBridge cost? Is there a free plan?

You get 100 free minutes on sign-up, no credit card. Among the rates: WebSocket streaming 0.0045 €/min, inbound on your number 0.0050 €/min, BYOC outbound on your own numbers 0.0050 €/min, number fee 3.00 €/mo; channel plans start at 18.50 €/mo for 1 simultaneous channel. Inbound, streaming and BYOC outbound are billed per minute, prepaid; other outbound calls are billed per second.

Switch on the bridge.
100 free minutes.

No credit card. SDKs on GitHub. Documentation in four languages.

Create a free account