Voice API Technical Specifications

Protocol & Transport

EnableX Voice is built on industry-standard signalling and media protocols. Whether you are placing a call through the REST API or connecting via SIP, the platform uses well-established RFCs that your existing telephony infrastructure is already familiar with.

Audio Codecs

EnableX Voice supports a range of audio codecs. G.711 (both A-law and U-law variants) is the preferred codec for standard telephony — it offers near-lossless audio quality at a fixed 64 kbps bitrate and is natively supported by virtually all SIP endpoints and PSTN gateways.

Security

The EnableX Voice platform applies encryption at every layer — from the REST API down to the media stream. Here is how each surface is protected.

Connectivity

EnableX Voice supports both SIP-based connectivity for existing telephony systems and REST API access for application-level integration. IP allowlisting is available for accounts that require firewall-controlled access.

Recording

EnableX Voice supports call recording in two modes. Recordings are stored on the EnableX platform and are accessible via the API after the call ends. See the Call Recording page for the full API reference.

Scalability

The EnableX Voice platform is designed to scale with your application load. Capacity scales horizontally across the infrastructure — there is no single-node bottleneck for call handling or API throughput.

Text-to-Speech (TTS)

EnableX Voice integrates with multiple TTS providers. You select a provider at the API level when using the play action with a text field. Each provider has different strengths — from ultra-low latency to deep regional language coverage — so the right choice depends on your application's geography, voice quality requirements, and latency budget.

The table below summarises each available provider.

Speech-to-Text (STT)

EnableX Voice supports multiple STT engines for real-time transcription during calls. The engine you choose affects transcription accuracy, latency, language coverage, and your ability to handle domain-specific vocabulary. Match the provider to your use case: latency-sensitive voice bots benefit from low-latency engines, while post-call analytics can prioritise accuracy over speed.

Custom Voice Prompts

Instead of generating speech via TTS on every call, you can upload pre-recorded audio files as named prompts and reference them by name in your Voice API responses. This gives you full control over audio quality, timing, and delivery — and avoids TTS generation latency for frequently played messages.

Format Requirements

Custom prompts must be uploaded as WAV files. The platform does not accept MP3 or other formats for prompt uploads. Ensure your audio is recorded at a sample rate appropriate for telephony (typically 8 kHz mono for PSTN calls, or 16 kHz for wideband).

Using a Prompt in the Voice API

Once a prompt is uploaded via the Portal or API, you reference it by its prompt_name in the play action of your Voice API response. The platform streams the pre-recorded audio to the caller instead of invoking a TTS engine.

{
  "play": {
    "prompt_name": "welcome_message"
  }
}

The prompt_name value must exactly match the name assigned when the prompt was uploaded. Prompt names are case-sensitive.