---
id: soniox
version: "2.9.0"
license: Apache-2.0
license_treatment: permissive
maintenance: active
---

# soniox — The official Python SDK for the Soniox API (STT, REST)

License: permissive · Maintenance: active · Downloads: 296.1K/mo

## What it is and what it does

Soniox is the official Python SDK for accessing Soniox's speech and audio APIs. It provides both synchronous and asynchronous clients that wrap REST and WebSocket endpoints for speech-to-text transcription, text-to-speech synthesis, voice cloning, and usage/cost reporting. The SDK handles authentication via API key, file uploads, model selection, and typed request/response objects through pydantic.

Typical workflows include transcribing audio files end-to-end in a single call, streaming realtime speech recognition or synthesis over WebSocket, cloning voices from reference audio for custom TTS, and querying daily spend and per-request logs. The package is actively maintained, supports Python 3.10 through 3.14, and has minimal dependencies (httpx, pydantic, websockets).

Use it for:

- Transcribe audio files or URLs to text in a single call with automatic cleanup.
- Stream live audio over WebSocket for low-latency speech-to-text in real-time applications.
- Generate speech from text with built-in or cloned voices, optionally streaming audio chunks as they arrive.
- Clone a custom voice from a reference audio clip and reuse it across TTS requests.
- Track API usage and costs per model and per request over a date range.
- Build async applications that handle concurrent STT/TTS operations without blocking.

## Worth the install?

AI-flagged interpretation of the facts above — verify before relying.

Python SDK for the Soniox speech-to-text and text-to-speech API, offering sync and async clients for REST and realtime WebSocket access, plus voice cloning and usage tracking.

Yes. The SDK is actively maintained, has low install friction, no known vulnerabilities, and a permissive license. Install it if you need to integrate Soniox's STT or TTS APIs into a Python application. Avoid Python 3.13.6 specifically due to a documented ssl regression; any other 3.10–3.14 version is safe.

## Install

pip install soniox
uv add soniox
poetry add soniox

## Installing soniox

Before you install: Low friction install with three lightweight runtime dependencies (httpx, pydantic, websockets). Active maintenance—last commit 2026-08-11, released 3 days ago. Supports modern Python versions (3.10–3.14). Note: avoid Python 3.13.6 due to a CPython ssl regression that hangs realtime STT/TTS.

License in practice: Apache-2.0 (permissive). No restrictions on commercial use, modification, or redistribution; attribution required.

Quickstart:

from soniox import SonioxClient

client = SonioxClient()
transcript = client.stt.transcribe_and_wait_with_tokens(
    file="path/to/audio.mp3",
    delete_after=True,
)
print(transcript.text)
client.close()

Requires SONIOX_API_KEY environment variable or explicit API key passed to client. Avoid Python 3.13.6 for realtime operations.

Verify before relying:

- Pricing model and rate limits for STT/TTS operations beyond daily cost tracking.
- Whether voice cloning quality and speed meet production requirements for your use case.
- Latency characteristics of realtime WebSocket endpoints under load.

## Package facts

- License: Apache-2.0 (permissive)
- Python support: supports_current
- Install friction: low
- Maintenance: active
- Downloads: 296.1K/month (top 15,000 on PyPI)
- Known vulnerabilities: none known

## Tags

speech to text python sdk, text to speech api client, stt tts realtime streaming, voice cloning sdk, soniox python client, async speech recognition, websocket audio streaming, speech-recognition, text-to-speech, realtime-streaming

[View on SkillFed](https://skillfed.io/packages/soniox) · [View on PyPI](https://pypi.org/project/soniox/)