Release April 2023

Streaming API Migration

Background

The existing streaming API endpoints /burst, /face, /prosody, /language, and /multi have been deprecated as of April 1, 2023. Existing users of these endpoints should migrate to the new /models endpoint which offers all of the same functionality plus some additional features.

Starting June 1, 2023 support for the endpoints listed above will be discontinued.

Please see the API Reference documentation for specific details about the /models endpoint.

Affected Endpoints

Endpoints being deprecated:

• /burst
• /face
• /prosody
• /language
• /multi

New Features

For details all new API features please check out the API Reference documentation.

• Run multiple models on the same input data simply by configuring multiple models in the same payload.
• Send raw text data without base64 encoding it using the raw_text parameter.
• Specify how much audio context is retained over the full streaming session with stream_window_ms and reset_stream.
• All detections across video and audio are now returned with the timestamps for each detection.

Migration Timeline

Updating your API calls

Individual model endpoints (deprecated)

// wss://api.hume.ai/v0/stream/face?apikey={{apiKey}}
{
  "image": "<encoded-media>"
}

Multi endpoint (deprecated)

// wss://api.hume.ai/v0/stream/multi?apikey={{apiKey}}
{
  "models": ["face"],
  "data": "<encoded-media>"
}

Models endpoint (new!)

// wss://api.hume.ai/v0/stream/models?apikey={{apiKey}}
{
  "models": {
    "face": {}
  },
  "data": "<encoded-media>"
}

Using the Python SDK

Versions of the Hume Python SDK 0.1.7 and earlier will no longer be supported as of June 1, 2023.