Speech-to-Text with Voice API and TeXML

​

Introduction

​

Video Tutorial

​

Supported engines

• Google (default) - Google speech-to-text engine that offers additional features like interim results.
• Telnyx - In-house Telnyx speech-to-text engine with significantly better transcription accuracy and lower latency.
• Deepgram - Deepgram speech-to-text engine with 3 models (nova-2, nova-3 and flux) that can be set using transcription_model setting.
• Azure - Azure speech-to-text engine with a strong support for multiple languages and accents.
• xAI - xAI Grok STT engine with the xai/grok-stt model.
• AssemblyAI - AssemblyAI Universal-3.6 Pro engine with the assemblyai/universal-3-6-pro model ID. The legacy aliases assemblyai/universal-3-5-pro and assemblyai/universal-streaming also run Universal-3.6 Pro.
• Speechmatics - Speechmatics real-time engine with the speechmatics/standard model (high accuracy with multilingual and bilingual language packs) or speechmatics/linden-1 (preview, vendor-hosted, 67 language codes, no auto-detect; defaults to en).
• Soniox - Soniox real-time engine with the soniox/stt-rt-v4 model. Automatic language detection with interim results and endpointing support.
• Parakeet - Self-hosted NVIDIA Parakeet engine with the nvidia/parakeet-v3 (automatic multilingual detection) and omi-health/omi-med-stt-v1 (English-only medical) models. Final transcripts only.
• Cohere - Self-hosted Cohere engine with the cohere/ar-stt model. Arabic and English transcription only; requires an explicit ar or en language — no auto-detect.
• Reson8 - Reson8 turn-based engine with the reson8/turns model. Automatic language detection across 10 European languages; transcripts are delivered per turn of speech.
• Humain - Humain real-time engine with the humain/realtime model. Arabic and English transcription with Arabic/English code-switching (codeswitch or auto).

​

Voice API

curl -i -X POST \
'https://api.telnyx.com/v2/calls/{call_control_id}/actions/transcription_start' \
-H 'Authorization: Bearer YOUR_API_KEY' \
-H 'Content-Type: application/json' \
-d '{
    "language": "en",
    "client_state": "aGF2ZSBhIG5pY2UgZGF5ID1d",
    "command_id": "891510ac-f3e4-11e8-af5b-de00688a4901",
    "transcription_engine": "Google/Telnyx/Deepgram/Azure/xAI/AssemblyAI/Speechmatics/Soniox/Parakeet/Cohere/Reson8/Humain"
}'

"data": {
   "record_type": "event",
   "event_type": "call.transcription",
   "id": "0ccc7b54-4df3-4bca-a65a-3da1ecc777f0",
   "occurred_at": "2018-02-02T22:25:27.521992Z",
   "payload": {
        "call_control_id":           "v2:7subYr8fLrXmaAXm8egeAMpoSJ72J3SGPUuome81-hQuaKRf9b7hKA",
        "call_leg_id": "5ca81340-5beb-11eb-ae45-02420a0f8b69",
        "call_session_id": "5ca81eee-5beb-11eb-ba6c-02420a0f8b69",
        "client_state": null,
        "connection_id": "1240401930086254526",
        "transcription_data": {
        "confidence": 0.977219,
            "is_final": true,
            "transcript": "hello this is a test speech"
         }
    }
}

​

TeXML

<?xml version="1.0" encoding="UTF-8"?>
<Response>
  <Start>
    <Transcription language="en" transcriptionCallback="/transcription" transcriptionEngine=”Telnyx” />
  </Start>
</Response>

%{
    "AccountSid" : "6d547b4f-993a-4e87-b95c-2d9460b3824b",
    "CallSid" : "v3:xIscDTsILHoErg5d4BfFWITg7vHmTvTRm-4YEeOgrwESDQsDWQNxvw",
    "CallSidLegacy" : "v3:xIscDTsILHoErg5d4BfFWITg7vHmTvTRm-4YEeOgrwESDQsDWQNxvw",
    "Confidence" : "0.9822598695755005",
    "ConnectionId" : "1614262910593271041",
    "From" : "+18727726007",
    "IsFinal" : "true",
    "To" : "+48664087895",
    "Transcript" : "let's hear some music"
}