Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.getpatter.com/llms.txt

Use this file to discover all available pages before exploring further.

​
Telnyx STT

TelnyxSTT is a streaming speech-to-text provider that talks to Telnyx’s wss://api.telnyx.com/v2/speech-to-text/transcription endpoint. It implements Patter’s pipeline-mode STTProvider interface — drop it in anywhere DeepgramSTT or AssemblyAISTT would go.

​
Why use it

If you’re already on Telnyx for telephony (see Carrier — Telnyx), you can keep STT on the same vendor:
  • One bill, one credential. No extra Deepgram / AssemblyAI / Speechmatics account to manage.
  • On-network audio path. Caller audio arrives at Telnyx’s edge for the carrier hop; routing it to Telnyx STT from the same edge keeps the audio inside their backbone instead of egressing to a third-party STT.
  • Pluggable backend. Telnyx fronts four transcription engines (telnyx, google, deepgram, azure) behind a single API — pick whichever quality / language profile suits the call without changing wire protocol.

​
Install

TelnyxSTT ships in the core getpatter package — no extras needed: 
pip install getpatter

​
Quickstart

import asyncio
from getpatter import Patter, Telnyx, ElevenLabsTTS
from getpatter.providers.telnyx_stt import TelnyxSTT, TelnyxTranscriptionEngine

phone = Patter(carrier=Telnyx(), phone_number="+15550001234")   # TELNYX_API_KEY from env

agent = phone.agent(
    stt=TelnyxSTT(api_key="KEY...", language="en"),             # or transcription_engine="deepgram"
    tts=ElevenLabsTTS.for_telnyx(),                             # ELEVENLABS_API_KEY from env
    system_prompt="You are a helpful assistant.",
)

asyncio.run(phone.serve(agent))

​
Constructor

TelnyxSTT(
    api_key: str,
    language: str = "en",
    *,
    transcription_engine: TelnyxTranscriptionEngine | str = TelnyxTranscriptionEngine.TELNYX,
    sample_rate: TelnyxSTTSampleRate | int = TelnyxSTTSampleRate.HZ_16000,
    base_url: str = TELNYX_STT_WS_URL,
    session: aiohttp.ClientSession | None = None,
)
ParameterDefaultDescription
api_key / apiKey— requiredTelnyx API v2 key (Bearer token).
language"en"ISO 639-1 language code (e.g. "en", "es", "fr").
transcription_engine / transcriptionEngine"telnyx"Backend engine. One of "telnyx", "google", "deepgram", "azure".
sample_rate / sampleRate16000PCM input sample rate in Hz. Patter sends a streaming WAV header on the first chunk.
base_url / baseUrlTelnyx prod URLOverride for testing.

​
Engines and sample rates

The Python SDK exposes the supported values as enums for editor autocomplete: 
from getpatter import (
    TelnyxTranscriptionEngine,    # "telnyx" | "google" | "deepgram" | "azure"
    TelnyxSTTSampleRate,          # 8000 | 16000 | 24000
    TelnyxSTTInputFormat,         # "wav"
)
Sample rateUse case
8000Match Twilio Media Streams μ-law without resampling.
16000 (default)Telnyx default media-streaming format (L16 PCM).
24000Higher-fidelity inputs (e.g. browser RTC).

​
Pricing

Telnyx STT pricing varies by transcription engine and is bundled with your Telnyx plan — see the Telnyx STT pricing page. Patter does not charge an STT line item for TelnyxSTT; the cost shows up on your Telnyx invoice alongside the carrier minutes. See Metrics for how to register a custom rate via Patter(pricing={...}) if you want it surfaced in the dashboard.

​
Low-level usage

from getpatter.providers.telnyx_stt import TelnyxSTT

stt = TelnyxSTT(api_key="KEY...", language="en", transcription_engine="deepgram")
await stt.connect()
await stt.send_audio(pcm16_chunk)                 # 16 kHz PCM s16le mono
async for t in stt.receive_transcripts():
    print(t.text, t.is_final, t.confidence)
await stt.close()

​
See also