Speechmatics Real-Time API Client

1.0.0 verified Mon Apr 27 auth: no python

Official Python client for Speechmatics real-time speech-to-text API. Version 1.0.0, supports Python >=3.9. Actively maintained.

pip install speechmatics-rt

Common errors

error ModuleNotFoundError: No module named 'speechmatics' ↓

cause Wrong import path; the package installs as `speechmatics_rt` not `speechmatics`.

fix

Install the correct package: pip install speechmatics-rt then import as from speechmatics_rt import WebsocketsClient.

error AttributeError: module 'speechmatics_rt' has no attribute 'WebsocketsClient' ↓

cause Forgetting to import the class from the module.

fix

Import correctly: from speechmatics_rt import WebsocketsClient.

error RuntimeError: Connection closed with status 401 ↓

cause Invalid or missing API key.

fix

Set SM_API_KEY environment variable or pass a valid API key to WebsocketsClient.

Warnings

gotcha The library uses underscore in its Python module name: `speechmatics_rt`, not `speechmatics-rt`. ↓

fix Use `from speechmatics_rt import WebsocketsClient`.

gotcha The `run` method is blocking; for non-blocking use `run_async`. ↓

fix Call `client.run_async(audiostream, conf)` instead of `client.run(audiostream, conf)` for async operation.

breaking Version 1.0.0 removed support for Python <3.9. ↓

fix Upgrade to Python 3.9 or later.

Imports

WebsocketsClient

from speechmatics_rt import WebsocketsClient

Note underscore in module name.

Quickstart

Example of connecting and transcribing a local audio file.

from speechmatics_rt import WebsocketsClient
import os

# Obtain an API key from Speechmatics
api_key = os.environ.get('SM_API_KEY', '')

client = WebsocketsClient(api_key)

# Define transcription configuration
conf = {
    "type": "transcription",
    "transcription_config": {
        "language": "en",
        "operating_point": "enhanced"
    }
}

# Open a connection and send audio file
def on_transcription(message):
    print(message)

client.add_listener('transcription', on_transcription)

with open('audio.wav', 'rb') as audio:
    client.run(audiostream, conf)